/robowaifu/ - DIY Robot Wives

Advancing robotics to a point where anime catgrill meidos in tiny miniskirts are a reality.

Server and LynxChan upgrade done. Canary update and a general address soon. -r

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB

More

(used to delete files and postings)


Remember, /agdg/ 's Game Jam is still going on. Go Team! :D


Prototypes and failures Robowaifu Technician 09/18/2019 (Wed) 11:51:30 No.418
Post your prototypes and failures.

I'll start with a failed mechanism, may we build upon each others ideas.
>>418
what was the intended function anon? did you ever solve it?
>>957
>Did solve it?
Getting closer, feel free to help.

-pic related are what I've been drawing for a leg mechanism where the foot is always parallel to the trunk, eliminating the need for strong ankle servos. Limits her ability to use them for sexual though, a worthy trade-off to me-
>>958
>feel free to help
I think someone already did anon.
>related
[[807

This simple mechanism already performs the specification you indicated. Just buy one and take it apart if you need to.
>>959
I forgot to mention all leg servos are localized within her hips to lower [the] inertial load [of] her legs on her servos.
>>960
Good idea, but I think you'll discover if you actually take my advice that [the swing-arm] mechanism is perfectly amenable to that design approach. You can think of the whole think upside down from a usual desk arm lamp with the weighted base as representing the mass of your servos. Still works identically. Don't forget to add in appropriately-scaled springs anon.
>>962
Mind giving us more written explanations of your design anon?
>>963
It's system that works like clocks, there's a motor connected directly to the rod which connecting its shaft to the shaft of the next joint. There's a concentric gear around the shafts which moves according to belts which transfer another motors motion to the final fixed gear.
Single motor walking mechanism, made in color to help illustrate its function. Capable of many different angles and speeds depending on lengths.
Simple single motor leg design.
Simplest single motor leg design I can come up with.
>>964
>>965
>>966
>>967
Have you created any physical models to experiment with your designs yet anon? I haven't played with them for a while, but I was thinking you could use K'NEX to prototype with. Maybe 3D-printing the parts? They look interesting BTW. Good luck.
Like I mentioned elsewhere, I'm starting with basic bots, working my way to more complex machines. From left to right – line follower, dancing otto, walking hexapod, obstacle avoidance, bi-directional sumobot, and expanding sumobot. So far only bots 1 and 4 work as intended. The dancing bot version I made is trimmed to use as minimum plastic as possible… its legs keep falling off (no wonder there's a more expensive version that uses metal gear connections). I still couldn't get the hexapod to work since the servo libraries I've tried don't work on an attiny85. I'm currently working on the sumobots which I hope will be able to battle each other.

I'm starting to get the hang of designing and printing parts that actually fit the electronics and any existing modules I have. I start by looking for easy-looking pieces on thingiverse then customizing them further then finally designing pieces from scratch.

After the sumobots I plan to make a toy grade 3D printed RC car that uses phone bluetooth + arduino control as well as an RC tank. Then I think I'll finally be able to make some waifubots, including the 1/8 scale rachnera-san and 1/3 scale wrestler bot.
>>969
>rachnera-san
</monster/ pls.

Wow, looks like you have been been busy anon! Good luck.
>>967
>>968
Tested physically, works but, not that well. Mostly just moves the lower leg segment
>>971
Maybe it would be cheaper/faster to prototype initial designs as cardboard cutouts instead?
Tried making a 10-spoke speed encoder wheel that fits the Tamiya 4WD tires, its too brittle and uneven. My plan is to have the optical speed encoder just arch over the main wheels while they spin, instead of trying to read a separate encoder disc. I will try 5 spokes and wont bother with concave shape, I'll just have the holes flat and square. The diameter with the Tamiya wheels is 28.5mm, one revolution is 9cm, so two revolutions or 10 clicks is exactly one maze block traversed.

>>969
Repost but the sumobots in action:
streamable.com/f2nqs

In the horizontal pic the bots in order of completion are 1,3,4,2,6,5.

Future bots are all waifu types, using some variation of the Fluffytail design and/or 3D doll pieces:

7. Micromouse (has its own current thread)
8. Cattymouse(larger micromouse with stepper motors)
9. Penguingirl balancebot(bluetooth control)
10. Kiwigirl balancebot(using 2.4Ghz RC radio and stepper motors)
11. Fluffy Rachnera-san(12-servo quadruped)
12. Fluffy Wresler(at least 14 metal gear servos)
13. Panzergirl waifutank (torso on tracked tank chassis)
14. Panzergirl devastator (torso on mecanum wheels). 13 and 14 will dress up like pic related.

Unfortunately neither of the future bots will stand taller than 40cm, I do not have the mechanical know-how to design for high torque, high weight applications.

I've noticed in related news, consumer home robots are closing shop: the makers of Kuri, Jibo, Cozmo, Vector. Most of the AI of those robots are in private servers so the bots will essentially die unless the AI is open-sourced.

Apparently the only successful "robots" are the Roomba, Siri, Alexa, Google Home etc. Either single-function tools or subsidized spyware. That is not good.

We must change that. Right now all I can do is work on creating a standard list of serial commands so that all my future waifubots can be controlled through a single universal arduino-compatible controller - either a physical controller that uses bluetooth or 2.4Ghz RC frequencies, or an app. Later waifus will have audio and video processing so I'm hoping I can offload that to a desktop computer.

I'm just sharing this infodump since I hope even though our paths may be plagued by a thousand PLA carcasses, there is a clear vision of what progress should look like, for me at least.
>>973
>13 and 14 will dress up like pic related.
<robowaifu qt tankgrill enforcers
nice.

As for robot shops closing, it's neither unusual nor surprising for a nascent industry to have churn. One great thing that we have on our side is that the Wright Brothers/Henry Ford style home garage operations are a real possibility [for this industry atm], as you and others are proving. We certainly don't want the botnets to be the only possibility for anons to turn to in the future though, so we have to stay focused. Fortunately we still have a few years before they manage to make it illegal to create your own waifu but have to use theirs instead. But it's definitely a race. If we make positive progress first and it becomes widely popular among hobbyists the world over, then we may be able to forestall such a move by the botnet vendors such as Jewgle & Amazon.

There are some command protocol works already out there, have you had a look at any of them?

Keep up the good work anon!

Sorry for off-topic OP
>>974
Thanks man. It is indeed the only industry I can think of that is truly still in the garage phase. Actual laws regulating humanoid robots don't exist yet outside of science fiction stories, so this is still a window of opportunity before corporate takes over.

>There are some command protocol works already out there, have you had a look at any of them?

I'm taking a look at various arduino smartcar bluetooth controller apps for Android, will try to find some common string between them so you can just download any one of them to control my bots.
>>975
>so this is still a window of opportunity before (((corporate))) takes over.
Exactly. But no one should just relax because there's a gap for now, and assume they just won't try to stop it. They will. We just have to move faster so there will simply be far too broad access to basic robowaifus before they can stonewall it.
Open file (1.72 MB 3264x2448 IMG_20200723_184329.jpg)
Open file (1.58 MB 3264x2448 IMG_20200723_184543.jpg)
Open file (1.70 MB 2448x3264 IMG_20200723_184407.jpg)
My biggest failure in general was getting distracted and frustrated because of problems like computer failures and all kinds of things aka life. I need to avoid mind blockades, like doing sth only when I archived something else first (which doesn't happen then). Good example might be, not buying a 3d printer until one has learned CAD. Then not being motivated enough to learn it, which wouldn't be a problem if one had bought the printer first. My biggest failure in so far in regard to outcome was overestimating the power of magnets, and therefore believing I could build a simple muscle based on that. I wanted to have piece of metal attached to a nylon string, getting dragged through a flexible tube. Somehow with magnets lined up outside of the muscle. I found out small pieces of metal cant hold much, and a bit of plastic isolates it quite good. 🤦🏼‍♂️ The other failure maybe isn't one, it's rather a bit of prototyping and testing. Though it failed in so far as I got stuck. Since I couldn't wait for my printer, which I want to buy after I moved, I started prototyping molds for thighs with pulpe aka pappmache and plaster (pics realated). However, I would've needed liquid foam at some point, but I had difficulties to buy that where I live, so I stopped working on it. The approach isn't really good anyways, bc I wont be able to mirror two legs to make them fit together. I know that, though I think it might motivate me to look into building joints, using electronics to move them a bit and so on.... Better than doing nothing!
>>4429 >Better than doing nothing! 1000% better, at least! :^) Actually, laying up laminates (like pappmache) is probably a good idea fundamentally. I'd suggest something like wheat paste for the adhesive instead of plaster to keep the result lighter weight.
>>4430 In that case here it's about a mold not an exoskeleton, weight doesn't matter. But you're right about us needing to use a lot of layers for the outside. It's because sensors and cooling tubes.
>>4435 >But you're right about us needing to use a lot of layers >a lot I'm not sure I would put it just that way, after all from the ME aspect, lightweight is king. It's just that laminar structures can offer improved strength to weight ratios. For example, plywood is an a good type of laminar sheet with good properties (though generally probably too heavy for our specific use-case).
>>4437 We all have different approaches and varying priorities, so general assumtions don't work. I want to put in quite powerful motors and muscles, but bipedal walking isn't so important at the beginning. The first ones only need to bend their knees, spread their legs, and at some point turn around in bed on their own, ... The legs shouldn't be to light, bc this would feel strange if I would holding them up. Later iteration would be walking on all fours to the bathroom.
I spent all of yesterday trying to get a 3D model to use as a basis to start planning, and all of today trying to start it putting it together. Not only is my computer a potato and declines to work with the more complex meshes that I try to throw at it, but even taking a full minute between movement and actions I get crashes. Not that the crashes matter though, honestly I have no idea how to put the pieces of this mesh together. I've been trying to figure out which face is the top face, the bottom face, and even after going through a few hours of lectures and tutorials I am still no closer than I was a couple days ago despite the work put in. I think I have a fix, but I can't be sure. I have a spare computer I got for cheap when a buddy of mine upgraded with a better gpu, which should help, but also just making an approximated model by hand instead of trying to assemble an accurate model so that I may refine it as I go could make things run far smoother. Hard to tell though. A bit of a pain to have wasted a couple days though, I will admit.
>>4537 Fuck it, I'm going analog. Clay, physical references, a drawing board, and hand calculations only for the physical design. I'm not going to mass manufacture, so there is no need for computer designs. Clay and paper are far easier to manipulate and measure with.
>>4537 What program did you try?
>>4539 Fusion 360. Solid tool, but the files I used were fairly bulky, my computer is a potato, and I am not as experienced as I maybe should be, so I don't blame the program itself. I've had far more sculpting experience than CAD experience, and it seems to be working better, although not by much. At the very least, it is making me rethink my approach.
I looked into it myself, without trying, since I need to rebuild my laptop, only have a Raspi and tablet working right now, and these would really be to weak: >>4549
Open file (1.64 MB 3264x2448 IMG_20200801_211920.jpg)
I'm working on a similar spine than posted here: >>4563. It's just meant to be a test and early prototype, till I'll get my printer. The plastic is the lid/cap from water containers, I have some springs and two types of little spheres. Also some metal parts where a little fiber or line can run through, but I don't know their name. I can tell so far that it's annoying to build. I put these little metal parts into some plastic hole and in between the end of a spring so they would hold the spring in place. Sometimes it works, sometimes I just won't get these into the spring. Then again, they won't hold the spring well, but thery are to short to put a washer in. Also, it's generally just trial and error. I want to get the idea how a spine could work.
>>4653 That sounds great Anon, keep experimenting. I like that you are using spare parts laying around to test things out with. Doing that will help us all understand how you would go about creating a robowaifu inexpensively. Don't be too annoyed at difficulties, just be patient and methodical. Remember that in a real factory situation (even a tiny factory of your own devising) manufacturing almost always spends the resources needed to create helpful rigs that help assemble/manufacture things. Be thinking about some kinds of rigs that might help you in the future as you're figuring out how to assemble parts. >I want to get the idea how a spine could work. That's both a neat goal, and an important topic. We won't be able to have good robowaifus without having good spines Anon, so work hard at it! Thanks for sharing this here with us Anon.
Open file (97.11 KB 1532x948 eyeball.png)
Open file (42.65 KB 1532x948 avril-lips.png)
Open file (80.97 KB 1532x948 PM35S-Lever-01.png)
Open file (98.34 KB 1532x948 disc-holder.png)
I needed to a bit of a break from my software project(s). But I'm still on it! I finally started to get more serious about CAD, software will have to wait a little bit. I already tried out some stuff before, like beginning an eyeball were I wanted to put a camholder and magnets inside, also flexible lips out of TPU, possibly at some point an airmuscle with silicone rubber on the outside. I learned that I can import pictures into Solvespace to use it as pattern, which is exactly what I wanted. I used Avril Lavinge as pattern. However, I didn't get very far due lack of knowledge on how to use Solvespace and also made some conceptual mistakes. Now I finally got through some tutorials, being in a more relaxed mood and started to design a holder and lever for some small servo (PM35S) which I have for testing and maybe for an elbow. I also had the idea of using washers in a little case as holder for some axis. The holder can have an arm going into the disc which can rotate freely. It's a cheap and thin axle bearing if one has no ball bearing or there's no space for it. So, that's not much, but still something. I will post more about those projects in the correct threads as soon as I archived something more. Doing this feels often better than working on software, since it's easier to have some result and hold it in my hands (after printing), even if something is wrong with it.
>>8034 Glad to see you're still going strong Anon! It's good to take a break from time to time and focus on other things. 'Variety is the spice of life' they say. I hope you succeed at the designs. I really like the fact you're trying to find extremely economical approaches to designs like the washer-as-axle-bearing. This sort of rigging is what will enable many more men eventually to find ways to create robowaifus on their own. I suppose if you put some kind of seal enclosure around the whole thing, then you could put machine grease into it for durability. I hope you come up with lots of new designs this year Anon! :^)
>>8034 Keep up the good work Anon. >>8038
Open file (87.40 KB 1479x968 lower_arm_V1a.png)
Open file (40.30 KB 1512x967 nose-lower-part-V1.png)
Open file (69.32 KB 1389x908 tube_connector_V1.png)
Open file (91.19 KB 1920x1016 lips_4_parts.png)
I'm preparing a print for a very simple forearm right now. Didn't start printing yet, we'll see how it's going. It's TPU, which is not easy to print. I want to build at least something, ASAP, to demonstrate some progress. However, this is not a forearm with motors or muscles, but a soft TPU+foam placeholder, which is supposed to be connected to a elbow with a little servo. More a visual improvement than anything else. From there, we'll see how it goes. I also started to design other parts, but it's not easy. The arms and legs of a human are not round, but more complex shaped, which makes it more difficult to get it right. Also, I need not only molds or some TPU enclosure, but also air muscles and pseudo muscles to fill the inside of the body.
>>8350 Good on you anon! The more functional parts such as joints, hinges, levers and mounting plates that we have, the better! Even placeholder CAD for servos and servo horns are useful and can save other designers time when they are planning their robowaifu parts.
>>8350 Oh cool. Glad to see some new design progress Anon. Yeah, I've often thought about all the void spaces within a robowaifu's outer shells and how can we utilize that volume.
Open file (195.76 KB 1504x1100 Imagepipe_10.jpg)
>>8350 I delayed my TPU forearm print bc my print didn't stick to the bed. I had gluestick on it, but heated the bed after the first layer. I guess this was the mistake, the PVA disintegrated. I will try again with a advanced design in a while. I need to come up with a solution for designing arm and leg shapes in CAD. The print would have cost 7-8€ and I want to have a better design before I drop so much fillament into it.
>>8371 After the fail with TPU I wanted to make a simpler print with a outer shell out of more expensive PLA and a inner core of cheaper PLA. To make the core bigger and the shell thinner, I changed the design. I tried different variants, but Solvespace always exports a mesh with a strange error. Maybe a bug, idk. Big F. Whatever, I might print the original model, still with a PLA core instead of foam, but bigger outer shell then I would like to have. Or, since I'm working on improving those designs anyways, maybe I won't print any of it now. I like printing parts, but it has to make some sense.
>>8372 That error looks like some kind of pole (the 3D mesh technical term). Any chance you might have some kind of vertices begin collapsed down together in your model?
>>8373 >being collapsed down*
>>8373 It's not a sculpt and I don't know much about vertices. It's rendered from a CAD file. It actually doesn't show up as error in a MeshLab (non-manfold edges).
>>8350 I'm still on it. Since I have these little PM35S motors, but they have no means to hold a arm, I have to work something out. Like mentioned here >>5034 I was on my way to build a whole construction around it, to hold the motor in the middle. I stopped this approach for now, and work on a disk attached to the motor and holding the lever which is attached to a rather complex washer. Note: Don't buy motors without mount for some lever or arm. >>8372 For now I only printed out the simplest version of the forearm shell in some cheap PLA, which is quite heavy bc I used to many perimeters. I might only use this as as stand to put hands on top of it and have them moving for tests. It's more or less just like a heavy vase with two perimeters, but without bottom. >>thighs I want to try now, to build legs and arms out of 4-8 parts, the problem is to align them. Not sure if that will ever succeed. Maybe it will be enough to print some structure out of TPU to hold other parts inside and will need some smoothing on the outside by adding fabric and silicone rubber.
>>8456 That's an interesting looking design Anon. I hope you get everything sorted to your satisfaction. One thing I'm learning for myself is that I have to start out on something literally as simple as I can, b/c that's usually just about as hard as I can successfully manage on my own. Once I solve one small step, then I usually back up and look at what I've managed to do that time and that often gives me guidance on where to go next. So, I just make dozens and dozens of little steps like that until I succeed. Takes some patience, but it usually works for me as long as I just don't quit. Keep at it Anon! :^)
>>8460 >just don't quit I won't. Though, I might soon be busy for more than one month with moving. I'll try to still do something in that time. After that, I'll have more space to set up workplaces for different areas which we need to cover. Doing CAD mostly works fine while listening to podcasts, talks on YouTube or audiobooks.
>>8464 I hope you have a successful move Anon. I should be moving myself in the next couple of months as well. I'm trying to get better at programming, and also thinking about our operating system/computer hardware choices to help make robowaifus safe, secure, and reliable. It's a big area to understand it all! Anyway, everyone helping everyone else here goes a long way towards us all moving forward together.
Open file (212.46 KB 1092x1104 Failing_towards_success.jpg)
>>8456 Just some upgrade: I managed to find a solution how to build a holder for that motor, but it took me a lot of trying around. Felt quite fine working on it, for a while, but towards the end I started to get more and more distracted by other things. Probably bc I had enough of it. I also assumed that this cheap little motor is going to work, but didn't look into it. Turned out, if it's going to work I'll need a special driver or setup with a H-Bridge. Whatever, I'm currently focused on learning and training CAD, so there's no rush. It was more about learning how to solve a problem and generally about dealing with that specific one. Turned out, I wasn't thinking enough about it before getting started. After trying complex designs, a rather simple solution works now. I'll upload the file soon, maybe later today.
>>8561 Actually, your persistence is laudable Anon. Just keep at it. >After trying complex designs, a rather simple solution works now. This isn't uncommon. Many designs can only achieve simplicity after the complex route has revealed the elegant route. Not only is this true in design and engineering, but in things like writing as well. There's a well-known quote written by the French mathematician and philosopher Blaise Pascal: >"I have made this longer than usual because I have not had time to make it shorter." Pretty much every piece of software I've ever cared enough about to spend enough time to pay attention to it's structural design has been through this basic process.
Open file (125.15 KB 1374x945 PM35S_tubholder_V2b.png)
Open file (482.86 KB 1535x2048 hamcat_mqq-28.jpg)
>>8561 I linked the upload here, in the skeleton thread >>8584 >>8568 Thanks for your encouraging words. But this was me scolding myself for not using my time better. I could do more. Next project: Early hamcat_mqq-style ball-joints. One thing I ask myself while starting: How round do these ball-joints need to be. Can they be a bit flat on one side? Pear-shape to reduce some material on the other side?
>>8585 Thanks for the pic of the hamcat_mqq Anon. That looks remarkably like the simplistic armature design I myself have been working towards. I may try to integrate your fittings/joints design in my own efforts. Thanks for your work.
>>8588 >hamcat_mqq more of his stuff is in the projects dump: >>7707
Open file (186.56 KB 1504x1262 Imagepipe_12.jpg)
>>8585 I was doing ball-joints today. Print-in-place was a difficulty, this means printing a mechanism at once, which then needs to break some supports of and being moveable. I got it working, by adding a hole to the bottom, so I could use a screwdriver to disconnect it from the ground of the cup. However, I'm still able to get the ball out and in again, which makes it pointless. It will most likely be better to print the parts separately, which makes them print cleaner. Currently it doesn't move smoothly.
>>8661 That's a neat idea about how to do mechanisms though. Just keep trying until you perfect things Anon. You get there! :^)
Open file (69.37 KB 1388x948 ball-joint-V4.png)
>>8662 I think I have to be careful with the print-in-place idea. It's not necessary and trying to make it work is eating most of my time with every prototype I design. Printing parts separate also means they are cleaner and therefore run smoother. Though, it's probably so, that we'll need to add something to the outer shell of the sphere (ball) anyways, to make it smooth when moving but also having more grip. I was thinking about silicone rubber, but not sure yet. However, this is one more reason to print the parts separate.
>>8690 I like the way you're coloring/rendering those parts in the diagrams. Makes them easier to understand actually.
Open file (186.40 KB 1433x1504 Imagepipe_13.jpg)
>>8698 Yeah, but I should have converted the last pic into a jpeg, 5MB is obscene, the jpeg has 200K or so. These pics are the actual models, just with everything hidden. They look like that when I'm working on them, just with additional lines and points. The pic here shows the difference in quality between print-in-place (middle) and separated (bottom). Don't mind the little gap in the bottom sphere, I'm printing with 15% infill. Also, the bottom sphere is bigger, bc I wanted to test if I would still get it into the cup if I increase the size of the sphere 10% (Answer: No).
>>8699 Actually to my eye, it seems you're already coming up with nice shapes. Put something like that through a resin printer and you would basically have production-ready parts ready to ship off to manufacturing. You're making great progress Anon. Please continue doing so! :^)
>>8699 BTW, Anon posted a video about an android leg that uses a big ball-and-socket joint for an ankle. >>8704
I was playing around with the idea of re-creating the mechanism of the InMoov neck accuator after my posting here >>8718. I didn't look at the details, just trying it on my own. The first part (pic 1) wasn't printable, even with these supports. It's to steep. Number two also wasn't that great. Don't know how to describe it well: The outer rails were too thin, since it's only the edges that turn. I played around with some changes, but nothing made sense. I was able to print it, but it was quite dirty. The third try was more similar to the original design. First I got it wrong again, but I'm sure V3 would work. I didn't really try to print it yet, but it looks the same than the original and I printed that already. The details matter, the outer lines have to be curvy. You can see it in the part I marked. Picture number four is the shoulder part for the doll >>8585. I didn't print it yet and it will need some adjustments till it's really finished. I partially disassembled my printer to exchange the heatbrake, so let's hope I won't break anything.
>>8760 This is actually really informative watching your progress and following your thinking Anon. You're already able to design a wide array of parts, it's pretty encouraging to see.
Open file (85.71 KB 1287x948 shoulder-joint-V2a-1.png)
Open file (88.75 KB 1287x948 shoulder-joint-V2b-1.png)
Open file (131.88 KB 1287x948 shoulder-joint-V3a-1.png)
Open file (95.47 KB 1287x948 shoulder-joint-V4a-1.png)
Open file (81.54 KB 1287x948 shoulder-joint-V4a-2.png)
>>8760 I had some trouble with my printer after finally replacing the heatbreak with a all-metal one. Also some other distractions. I did still work a little bit on a prototype for a alternative shoulder joint, which is not a ball-joint. The blue part is supposed to be constraint to the body, the green part can then slide sideways on it, but is also supposed to tilt up and down. For lifting the arms sideways. Though the later I'll have to figure out by printing prototypes, which I can't do right now. The current design probably wouldn't work. I need to figure out the details and maybe add another moving part as well. The grey disc is a placeholder for a metal disc with bearings, so a attached arm on it could move without friction. This is for lifting the arm in front and back of the body. I wanted to have a holder for the disk integrated into the design at first, but realized that I want my parts to be printable with no or minimal supports. I need one plane side for that. So I added a hole to the later version of the design, for adding the disc with a holder using a screw. I might integrate some metal bearings into the other parts later as well, but this should be optional. However, I decided to build a prototype at some time (shipping problems currently) which will use motors in the chest and strings to move the shoulder. The idea is, that the strings can then later be replaced by airmuscles or other ones, so the motors shouldn't be necessary anymore. Though some might still be part of the design as a fallback solution. Since there won't be a motor in the shoulder the "bone" itself can therefore be kept small and be surrounded by some soft material. So her shoulder can look feminin and be soft like a human one. Btw, I'll plan to upload the files soon and also opening a Mega account so I won't need to use Catbox all the time. For now, in that case the idea is more important than the current execution anyways, and the signal that I'm still operational as well.
>>9142 Btw, the empty spaces in the middle of the organic looking part in the shoulder joint are for separating that part into two parts, for printing. So each one has a plane foundation, which will be on the printbed. They could be screwed or glued together, or I might later add some more space for a layer of machined metal inside, so they would be more durable. I will need to add contacts into the design as well, since I want to keep bending cables to a minimum. The hard parts of the joints are supposed to transfer energy and data through the body. The sliding parts need at least two metal layers to transport current, data might use some form of light communication based on LEDs with different colors. I still have to watch quite some videos on human anatomy and also read a little bit about it. There will also need to be a lot of trial an error. Then, I also spend some time trying to create a human-like ribcage in CAD. Not with much success. I used pictures from some scanned parts which I downloaded from Thingyverse or a similar site. It turned out to be hard to do that in Solvespace, at least with my skills. However, I want to do as much as possible in CAD instead of scultping, because it's easier to manipulate these files to add something later. So I'll try again at some point, probably after improving my skills in using the program first.
>>9143 I also found a front part, which would fit quite well to the chest of a hard-shell robot, but might also be interesting to print in a flexible material. It's a booby plate armor design, inspired by The Mandalorian. https://www.thingiverse.com/thing:2888954 I'm going to add this to the skeleton and armature thread as well, but only together with some other pics on my other device. I mention it here solely because it might be useful for adding a female look to some ribcage design very fast. Also, because it's part of the next picture. I used Dia, to put some of my pictures of prototypes and pictures of patters I want to use together, into a kind of diagram of what I'm working on. So I can have a overview and decide where I want to try next to push something forward. The pictures are sorted to some extend, to be circa at the place where the parts would fit into a build of a body.
These are excellent posts Anon. Seeing your design progress is inspiring. I'm particularly glad to see you working on the shoulder-girdle and ribcage areas. I think about those a lot and did a small study on the design and kinematics of the shoulder-girdle back in school. It's an amazing design actually. Keep up the good work, and may you have good success at it! :^)
Open file (59.25 KB 1226x948 ribcage-V2b-pic05.png)
Open file (348.83 KB 1226x948 ribcage-V2b-pic04.png)
Open file (59.63 KB 1226x948 ribcage-V2b-pic03.png)
Open file (64.00 KB 1226x948 ribcage-V2b-pic02.png)
Open file (51.57 KB 1226x948 ribcage-V2b-pic01.png)
>>9144 So, I was watching videos on how to get a ribcage and shoulders right in CAD, because that's what worries me. I hope I'll try out Blender soon. However, after watching some video on how to draw a chest I got inspired. My problem might really be that I'm not trained in drawing people, and I found it's worth looking into that. I really wanted to do a ribcage ,build like a human one. Starting from drawing a chest I tried something else, what you can see on the pictures. One also shows the amount of complexity by showing all the lines. This is of course once again just crude prototyping which might not lead very far. Also, FYI this is meant to be printed in TPU when it's finished. So I might still put in some ribs on the outside (in PLA), but mainly for show to make them visible through the skin, without really having a full ribcage. The thick parts of the prototype would be made thinner or hollowed out to hold parts like small computers and such. Btw, here a no tiddies bc its not the outside, there would still go at least a a layer of power mesh with silicon on top of it. Also, it's far from finished, just wanted to give an update. The video was on a male ribcage anyways, so I have to feminize it anyways. I want our waifus to have a bit unrealistic waist to hip ratio (wasp-like), tbh. Like they would be wearing a corset. I also put some silicon on top of a small printed face of Sophie to check if it sticks. Even without thinning it, so that it goes better into all the cavities it actually does. Silicone is known to not stick very well to anything that silicone, it only sticks better by having a huge surface to hug onto. I'm thinking of making experiences with some silicone on plastic faces. This isn't really the my preferred way, since I would rater like to create a skull and then add soft silicone parts, but one step at the time. My printer is still not working, if you are wondering why I don't try to print some parts. I'm working on it, from time to time, but have to keep an eye on my frustration management.
>>9620 > it's far from finished, just wanted to give an update. That's fine Anon, I'm glad you did! It's enjoyable watching how you progress in your work to the final products. > I want our waifus to have a bit unrealistic waist to hip ratio (wasp-like), tbh. Like they would be wearing a corset. I think that's a really good idea from a character-appeal perspective. However, it may prove a little tricky that way from an engineering perspective. For example, Sophie Anon's challenges fitting all her new eye gear inside her head. Regardless, I'm sure you'll manage Anon. BTW, any idea what's wrong with your printer ATM?
>>9620 >My problem might really be that I'm not trained in drawing people, and I found it's worth looking into that. You are absolutely correct Anon. Life drawing is basically the key to being able to 'see' how anatomy works. There are a large number of drawing tutorials out there in many places on the Internet. Our webring's own /loomis/ has some too: https://anon.cafe/loomis/res/628.html#628 Good luck. It takes time, but you'll eventually get it if you just don't quit.
The important thing about >>9620 is that I realized that a really should work with subtraction in CAD. I had the wrong idea that I would need to keep my models ultra simple and also ideally not render something to then cut something of it away afterwards. Sense of perfection or the goal of keeping it simple can stand in one's way sometimes. Assembling of different files also works in Solvespace, but it is a pain to do with rather organic parts that don't really fit into each other, have to overlay partially, but still don't fit very well (see my try on a thigh in >>8456). >>9621 Don't want to discuss my printer problems here ( in this thread), that's why I didn't mention the specifics. I already have tutorials and advice, but thanks for asking. >>9622 Some of the /loomis/ examples are directly useful, thanks. I'll have to look into how to use wget or curl to get all the images. Many are more interesting for drawing then for creating a 3D model, but I can use some of them. Also, maybe I might want to learn drawing one day.
>>9631 >...Sense of perfection or the goal of keeping it simple can stand in one's way sometimes. I'll presume you are referring to the psychological state known as perfectionism in the first part here? While this is definitely a hindrance in many ways to a man's progress in whatever given field, I consider the latter to be practically a law for success. Conflating the two together is a fallacy. The former is a kind of obstinate drive to perfect one's perceived notion or ideal about a thing or a system. These perceptions are obviously, necessarily, a priori, incomplete. This pursuit is on the face of it misguided in a fundamental way, and will certainly lead to one's personal pain and suffering. Balance in all things, and knowing when to quit are two important characteristics of a man's growth to maturity. You'll never get things perfect in this life. Good enough is, well, good enough. :^) However, the latter 'law' of simplicity is the primary effective means ever discovered that lets men tackle the absolutely unchangeable nature of the world around us. It's one of the rather few means at our disposable in our ever-running 'battle' with The Second Law of Thermodynamics (commonly known as Entropy). https://www.grc.nasa.gov/WWW/k-12/airplane/thermo2.html https://en.wikipedia.org/wiki/Second_law_of_thermodynamics The insulting mnemonic phrase 'Keep It Simple Stupid', while a good admonition, is actually a blatant misnomer. Only the 'stupid' man doesn't work to keep things simple in their lives. The intelligent man works hard to do so. This should be instead: Keep It Simple, Smartguy When you're dealing with the monumental complexity of devising a working, IRL, autonomous gynoid robotic companion -- which we as a team here on /robowaifu/ are endeavoring to -- then keeping it simple is literally the only possible pathway for us to eventual success. >tl;dr Put simply, all things around us tend to disorder. "Keep it simple" is a good way to keep this manageable, and maintain both order and sanity.
>>9631 >Some of the /loomis/ examples are directly useful, thanks. I'll have to look into how to use wget or curl to get all the images. I can look into compiling an archive of the whole board and posting it for you somewhere Anon. Might also turn out to be a good resource for more than just /robowaifu/'s use tbh.
Open file (110.08 KB 1424x948 ribcage-V2f-01.png)
Open file (97.61 KB 1424x948 ribcage-V2f-02.png)
Open file (42.05 KB 1424x948 Genera_hand-V1a.png)
Open file (50.73 KB 1424x948 Genera_hand-V1b.png)
Open file (28.98 KB 1424x948 Genera_hand-V1b-extr.png)
I didn't really archive much to report recently. I worked a bit on the chest, to put in spaces for ribs in another material and was also starting to work on a long term project of building the bones of a hand out of layers, which could then be PCBs (electronics) with sensors, plastics for the form and metal parts for strength. I also got some more stuff from AliExpress, but didn't really work with it. Some motors will arrive soon. I had some problems with my Raspi. I thought it was the software or my disc, since it got slower and slower, but it was the SD card which failed a while later. I finally looked into getting a new PC, which took me some time because I had to find out what to buy in the current situation and how to get it. I also finally worked on getting my Laptop running, which is weak, but better than my Raspi3. I had troubles to work on my design on my Raspi, I had to wait to often. I also had to look into not missing out too much on business opportunities, which I ignored in favor of my main dedication here. Not sure how much time I'll have during the next two month, but I'll work at least a little bit on my designs and with my new things. Though this might not lead to immediate results.
I also finally fixed this >>8372 >>8373 - In the configuration, the export chord tolerance (in mm) needs to be the same than chord tolerance (in percentage) which is also shown in mm.
>>10292 Ahh, good detective work.
>>10291 >I worked a bit on the chest, to put in spaces for ribs in another material and was also starting to work on a long term project of building the bones of a hand out of layers, which could then be PCBs (electronics) with sensors, plastics for the form and metal parts for strength. I've thought often about the field of Neuromorphic computing, specifically as it relates to designing/engineering robowaifus. Using structural and other ancillary parts embedded with sensors, batteries, microcontrollers, wiring, electronics parts, etc., right inside the structural and actuator components is not only very bio-mimetic in design essence, it also is very likely to help bring the extreme high-performance characteristics of neuromorphics to the table. For example, embedding temperature sensors directly within the finger bones, and also keeping the robowaifu's self-protection 'sense/react response cycle' to pull away from the heat, say, all 'short-circuited' locally right inside a simplified hand-local electronics/microcontroller/actuator system. This design approach can allow the response times for such a system to be very fast relative to a more traditional, hierarchically-networked command & control mechanisms. Basically, in a somewhat similar way to biologic adrenergic nervous system response mechanisms, you want to push the 'computation' for such a system out to the edges of the physical structure, and not be so dependent with always 'phoning home'' first to the higher-level computation systems of the robowaifu's 'mind'. This latter approach encompasses costly communications and other delays. Not that the signals wouldn't be sent on their way 'back up the chain' though. You definitely want the ability of higher-level control to override lower-level ones when needed. Forging ahead into dangerous environments to protect her master for instance, even when doing so conflicts with the most basic of self-preservation dictums. This round-trip would hopefully be completed within milliseconds (vs. the hopefully microseconds-level desired for pure local response times). My apologies for my probably confusing writing here Anon. This is a complicated topic and it's difficult for me to describe it concisely.
>>10297 As an additional thought on the specific example of HOT! PULL HAND AWAY IMMEDIATELY! example, the control devices could perhaps either open, or reuse, an emergency response communications channel up to further-up actuator systems in the robowaifu's skeletal chain. So for example, the hand-local would attempt to instantly flex fingers back, but then emergency-response channels can be opened to the wrist, elbow, shoulder, and torso actuators, all in a tiered-priority chain, to enable fully pulling the hand entirely away from the danger, same as we ourselves would do accidentally touching a hot iron for instance. Each of these chained-actuators would quickly add their own kinematic dynamic in the movement, and the effect would be propagating and progressive. The idea behind the 'emergency response' is that the higher-level analysis would be bypassed in a first-order response time, simply to quickly save the robowaifu from immediate damage.
>>10298 One additional thing that will need to be solved for this hypothetical situation. As we grow up, our entire physical being develops a kind of physical awareness that let's us intuitively discern where sensations are coming from in our body by mere touch, and usually more or less instantly. Vision and audio, for instance, are not needed to know you've just touched a /comfy/ soft blanket, or a cold ice cube spilled onto the counter. And not only do you recognize immediately these kinds of sensorial cues basically immediately, you also know where (to a first approximation) the touched item of interest is located, relative to your general body position. Again this is all instinctive to us, and happens 'automatically' with little attention needed for most cases to figure these things out. Back to the HOT! emergency response, the robowaifu's system will need some kind of touch location-finder mechanism so she knows instantly where the hot plate is, and which way to yank her hand back out of danger. If this isn't done accurately, she could make a clumsy move in the reaction, and possibly damage herself, you, or something else. Again, this is something we all develop instinctively as we grow up, but for us as designers and engineers we'll have to solve this kind of thing explicitly. I'd guess that a first-approximation approach would be to keep a general sense of all the items in her local body space area's surface normal. This should at the least give her the direction to quickly move out of the contact danger (ie, out along the surface normal of the object and away). This situational-awareness solution needs to account that this 'normal-map' of her environment is dynamic, as both she and the elements in her environment are potentially in motion with respect to each other. This is really quite a remarkable domain to tackle from a systems-engineering perspective. Now that I've been applying myself to consider some of the many things all needed, most other design & engineering endeavors seem rather boring to me now. :^)
>>10299 Also, once you check my digits another thing we might do is develop a sort of 'contact-pad volumetric triangulation' sensor model. The idea is you have many tiny pressure, etc. pads embedded into the robowaifu's 'skin'. Whenever she touched something, and approximation of it's shape (and by implication, it's surface normals) can be quickly simulated in her world model. For example, if 18 different pads on two of her fingertips all register a contact, then based on the kinematic/skeletal/etc body model simulation of her current physical position, then she can 'triangulate' the surface shape of that object at it's contact points with her fingers. Again, all instinctive for us...but for her will need to be explicitly worked out in advance by trial and error during design.
Open file (186.01 KB 1384x984 back_1st_try.png)
Open file (36.55 KB 1314x984 lower_arm_2st-try-001.png)
Open file (100.95 KB 1314x984 lower_arm_2st-try-000.png)
>>10291 I finally made a Mega account for my designs and testing results: https://mega.nz/folder/nv42EbzT#1MUblavva2UG6DukgD0moA Not that something happens to me and all would be gone. Also, SophieAnon shouldn't be the only one uploading design files. - They are all made in Solvespace for now. - Didn't add some copyright licence for now, because I wasn't able to decide yet. - I didn't add all my tests and failures, the other files are quite messy and I need to look through them first. - The structure of the folders might also change, it all just provisionally. - Didn't add stl files yet, I would first need to open the files and export them. You can do that yourself, but these are just prototypes anyways. - I want to try out more modelling with OpenSCAD, IceSL, Blender and maybe importing stl files into Solvespace as well. STL import is available in some some alpha version, but I didn't test it yet. - I want to try OpenSCAD and IceSL, because organic forms are just to difficult to do in CAD. But also because complex models would have bones which are dependent on each other. Think of the bones in a spine. I want changes applied if I change some part. I think more and more that this is only possible in code. Sculpting and mouse-based CAD might not work. - I also consider paying some professional to make a model of some face from an actress for me. Then, later an anime version of it. There seem to be cheap designers in poor countries offering that (3D/face modelling) as a service. At some point in the future I would like to build a AI generator to do this from photos, but I would need some examples anyways. - I will open my own thread as soon as I have more to show for. For now I call my project "Genera Project", but this might change. I want the project and my specific build (my robowaifu I'll build one day) to have different names.
>>12661 Glad to hear you're beginning archive your work online. I think SophieDev used more than one site in case his work was taken down by snowflake leftists b/c robowaifu wrongthink? Might be a good idea tbh. >For now I call my project "Genera Project" Good luck with Genera Project, Anon.
>>12661 I worked on a simplified model of a spine tonight. Well, one section of it. Kind of had a blockade about what to try next before I got started. Didn't know what I could and should try next. Can't work on the chest because it got to complex for my Raspi to handle, tried to install some Linux distro on my Laptop which gives me troubles. I might soon just install some random distro that works, and use it just for my CAD until I get my PC. I wanted to wait till I can use OpenScad and Blender on my laptop so I can try to design a good spine with that. I don't think I can do it in Solvespace, because it's to complex with parts dependent on each other. Decided eventually not to go for a perfect human-like s-curved spine for now and just try something out. It's supposed to work with twisted string actuators. Though, I don't plan to use them as sole actuators. It's just a incomplete model right now. I didn't add holes for the forward movement yet, because I might use other 'ribs' for that. I also could add them easily if I need to. The pink ball is just like some flexible ball out of plastic I bought for my cat. Maybe the size is good or maybe I have to find something else, for my prototype it will suffice since relation to other body parts won't matter. I also ordered some simple bags for 200 ml of liquid, which I hope to use in experiments for my overall design. I'm thinking of having horizontal layers in the waifu chest, containing bags out of fiber reinforced silicone, which then can be filled with air. Two would be on one layer with at least one normally closed valve in the middle. The valve would be in line with the spine, while the bags would be spread out on each side. This could provide some holding torque without e.g. having some motor providing torque all the time. The side with more air would keep the chest leaning in one direction, while it could also react very fast to high pressure or to some sensor, if the body is being forced to move by experiencing some force against itself (compliance in robotics). I also plan to try to combine this with horizontal layers of plastic sheet moved by magnets which would compress one side of such layer. So if the valve in the middle between two bags is open while the magnet compression on one side snaps, the air should go to the other side, moving the chest very fast (but without much force) to one side. Alternatively some pump might remove the air from the silicone bag on one side, while the air from the other side can't move over or maybe the amount of air is even additionally increased on that side, creating the same effect. Though, this version might be a bit slower, and make more noise. I also might have fixed my printer. Found some part which might belong into it, will need to find out about that and I didn't test it yet. Had no motivation to fix it for a while, because of frustration about the whole thing, so it stood around for a few months.
>>12882 This is very interesting Anon. Any chance you can piece together a render of a fully-assembled group of these into a complete spine? We'd certainly like to see it if you would for us please.
>>12887 I'll order some drone motors and then print it for testing. I have motors, but only one per type. They're very small, I wanted to use them with a cycloidal drive but failed in designing that in Solvespace.
Open file (179.70 KB 1920x1016 pelvis-01.png)
Open file (172.38 KB 1920x1016 pelvis-02.png)
>>12661 I just made a little test with a part (bottom/back) of the pelvis from >>12878. Not sure if it is taken from a male or female, there's not indicator of it. I didn't do much anyways, just slicing it in a crude way to make it possible to have a look at it and contemplate how it could be designed in some CAD program. It's for parts, two are always (more or less) the same as their mirror. So assuming the parts were simple enough now, designing two parts and then mirroring them would be enough. Related: >>12884
>>12905 *It consists of four parts, two are ...
>>12905 Nice work Anon. I think some degree of intentional biomimetics will be a key to success for us here.
>>12882 I went on working on it a little bit. Experimented with a chamfer to to make the edges less sharp, since the rope would rub there all the time. Making the design separable for printing worked and I added some holes for screws and nuts. However, I didn't print it yet so I can't know if screwing it together will work well, but it could be glued as well.
>>12923 Exciting to see your progress, both on this effort, and especially as a designer. You're becoming quite skilled at your work, Anon. Keep it up!
>>12926 >becoming quite skilled at your work Thanks, but it mostly isn't really difficult. Till it is. So far I failed in transfering Levi Janssen's approach to create good cycloidal disks to Solvespace. Yesterday, I also looked into how thick we can make a femure (upper leg) bone and how many LTO batteries could fit into each. It's also just a sketchy model so far. Six is the number I came up with, for now. Which gives us 10Ah with 14V or 20Ah with 28V. I guesstimate that three might fit into the lower leg bones.
>>12935 > guesstimate that three might fit into the lower leg bones. It's good thinking to distribute things around and utilize any available, latent, volumes. However, batteries and other massive (in the physics sense) items bring some additional caveats to the table for us as well. The concept of ' Thrown Weight ' (as used in race-car and other forms of dynamical-systems engineering) is something that directly affects our robowaifu engineering too. To wit; we need to keep as much of the mass centered around the pelvis volume as is feasible, and to not put it out in the extremities (head, limbs, hands, feet, etc) to the degree at all feasible. The further out from the center-of-gravity the mass goes, the exponentially worse the effect is in terms of kinematic and energy consumption, etc., penalties becomes. As you seem well-aware already, keeping the robowaifu's overall mass itself down is an important consideration in many ways. In like fashion, keeping the thrown-weight down is also very important to us. I'd suggest at the least not putting any batteries out past the elbows or knees to help alleviate the issue. Ideally, they would all go right into the pelvis/hip area.
>>12937 Thanks for your input, but please stop repeating that same standard text over and over again. I've read it 10 times already. I'm ignoring your conclusions on this topic anyways, because it doesn't work for me. Thank you. There's no way I'll put everything into the chest. First of all, when I'll lift her legs while she's lying they can't be to light. Then there's not enough space in the chest, especially not in a soft doll-like body. Also, she doesn't need to walk for miles. Five or ten minutes max is sufficient. While doing so, she mostly needs to lift her feet 1-2 cm of the ground. Last but not least, it would only matter if the downside for putting more batteries into her would be higher than to leave them out. Which most likely won't be the case if she's sitting, standing or lying down most of the time. I don't know how to calculate this but 20Ah aren't nothing. So her weight will be 30-40 kg, I guess. If it's less and I like it, then fine, if I won't like it I might even put more water into her internal tanks. On purpose, to make her heavier. I'm not MeidoDev >>11446, he's going for ultralight. Tbh, I might not do that thing with the additional water, but try to put more batteries, dielectric fluid or bigger motors into her design.
>>12949 lol. sorry, but you won't badger me by such talk. we're not women here on /robowaifu/. disprove my position with physics if you're so set against my point. I've encountered and proven sufficiently well the issue numerous times with racing designs. it's simple lever mechanics: take a 1m long stick and and 1kg weight. Perform two experiments with said setup: 1. Pivot the lever with the weight attached to it down right at the pivot point. 2. Pivot the lever with the weight attached to it out at the end of the lever. What else do I need say? That's the 'thrown-weight' issue in a nutshell. Exerting unnecessary energy through poor designs will be detrimental to a robowaifu in every way. Extra energy consumption, extra heat production, extra mass added to joints, connectors, fasteners, etc., to deal with the extra forces needed. Awkward motions in the kinematics. Slow motions in the kinematics. Costs go up in every area, financially and otherwise. Keep mass inboard, and you'll improve basically every area of your designs.
>>12952 >disprove my position with physics I didn't attack your arguments fundamentals based in physics. It's about repeating it at every opportunity again. I don't build a robot optimized for walking. Context, priorities and constraints matter. Feel free to build one without the batteries in the legs, and make them as light as empty bottles. But I wont.
>>12935 Here're my early and crude prototypes for the femure (upper leg bone) so far. The challenge is to make the part as huge as possible to hold as many batteries or other things (maybe motors), but make it so that the bone can't be felt from the outside and also leave space for soft tissue, the muscle movement and muscle mimicry on the outside. Currently I'm assuming that 9x9 or better 8x8 cm femure bone diameter might be the maximum, and that only in a rather tall or thick robowaifu. Transitional elements between parts with different sizes are still difficult to me to do in Solvespace, but here it wouldn't be visible anyways and it's just me trying something out anyways. Maybe such parts need to be sculpted in something like Blender and then imported. Also, blue was maybe a bad choice for the sample, looks rather sick. The smaller holes in the first part are also just an example. Those might be used for steel rods or these hollow metal tubes similar to broom sticks, to make the thing more resilient against shocks. But could also hold cables or water tubes/hoses, or guide some strings, though I think except of the cables these things would rater go into the soft part outside the bone. What's also missing in that model so far, are the heat exchangers for watercooling the batteries, which would need connectors to the water hoses inside or outside the bones. Later I will also need to think about how to transfer water from one part of the body to another, since all hoses tend to break when being bend very often.
>>12954 This is a nice design Anon. Looks like it would have a good strength-to-weight ratio too. Please keep going with it!
>>13091 Thanks, I will keep going but have various things to do. My motors for testing the spine also seem to be already in the country. I see often videos on problems with backlog in logistics but I'm still getting my stiff surprisingly fast.
Open file (3.04 MB 320x240 2019.mp4)
My 2019 prototype, it costs me 100 usd to build it w/o sex doll price. Now I am thinking to a whole new design but I would need a lot more money to do that, so have to wait. I used an Arduino Leonardo board and 4 or 5 servo motors. The mechanism was to use a lot of strings as tendon, similar as this prototype https://www.youtube.com/watch?v=0ZBD2tcKOU4
I used 500lb braided fishing line as string/tendon.
>>13541 Post her mechanics and circuits. Nice job
Open file (432.26 KB 1252x922 w.png)
>>13544 Thank you :p Since it's an old prototype, ive deleted everything except this. But here's how I did it globally: Red are servo and green is the fishing line. One fishing line per axis, thus 4 servos. The circuit is not that complex, It's similar as https://www.youtube.com/watch?v=ZySGP4AwGCY[Embed] The servo were initially 180 degree locked but ive unlocked to 360 degree. https://www.youtube.com/watch?v=JhHSXCLsN4k The servo I used is https://www.aliexpress.com/item/33010787343.html And its battery: https://www.aliexpress.com/item/1005003043860597.html Small and high torque, perfectly fit into my 100cm sex doll For the new prototype, I plan to work on a 128cm sex doll from now on, I will have more space for servos and electronic board. And I will switch to STM32 black pill board, I found its programming way more interesting. Basically I will deport the leg mechanism to the thigh instead and will reserve the chest for other further mechanism.
Oh no, I am realizing that the new design has a flaw, if it's designed like that it's the leg that will lift the whole body by its own, it's good but I also want the opposite. Now I am aware it's the abdominal muscles that are responsible of the lifting leg mechanism. Rip, thus I have to combine old and new design to have a full leg mechanism.
>>13552 Judging from her high mass and the use of a set doll, she's going to lay in bed and be a sex robot only, right? You did a good job on her skeleton and you're using good electronics to animate her. Looking forward to seeing how this goes. Also, I'd put her servos in her belly, then she won't have to move there mass.
Open file (40.15 KB 457x609 QlYjX1p.jpg)
>>13557 For now yes, I have in mind that we can only produce the first Humanoid with the help of sex industry since the amount of money it produces. So now she belong to the bed lole Thank you and yes, I am ignoring the belly too much, it's also due to the sex doll shape that has wasp waist. I should buy one with a normal shaped waist and also make my own skeleton to optimize the space even more.
Thanks for sharing. You might want to check out my Genera Platform (aka Genera Project) from time to time. I doesn't have a thread on it's own yet, and only making slow progress. It's also based on a endoskeleton instead of a shell. I'm currently posting my prototypes in the thread for prototypes and failures: >>418
Open file (65.30 KB 1298x984 holder_lower-v1a.png)
Open file (139.80 KB 1355x984 holder_twister-T1v1a-01.png)
Open file (81.31 KB 1298x984 holder_twister-T2v1a.png)
Open file (68.50 KB 1298x984 holder_upper-v1b.png)
>>12954 So, I uploaded some new files to my Mega account, which I linked here: >>12661 I got my JMT FN15844 BLDC motors which I want to use, for testing at least. For my spine and for some arm. I designed some files to add other parts to each of them. One way is using twisted string actuators, but I now hopefully found a way to finally get some cycloidal disks as well. This is all sketchy for now, I didn't print not test them yet. Just wanted to give some update.
>>13704 Still using Solvespaace? Your designs are getting really good considering the software limitations. Hope to see them irl soon
Open file (316.22 KB 1298x984 Waifu_full_body.png)
>>13704 I also created some way to create a diagram in Solvespace to have an overview over the system. I just realized it also needs some minor fine work till it's finished and can be used by others. I want it to look like in the last picture, so without the green squares, but then with the added parts and pointers. It is meant to be used by others to make such a diagram, but also to show off my design approach in a version of it. Like in picture one. The second picture is a well known pattern for drawing anime waifus which I uploaded already somewhere else. I used that to make the female form in the diagram, which should be usable by others for their designs as well.
>>13709 I was also working on some human-like knee. Just a test or feasibility study. Also unfinished. BTW, I decided to upload my files faster, but if I didn't even rename the internal parts properly then the files are in a subfolder named 'sketchy-in-progress' or so. Those will need some workover but they are now there as a backup. >>13708 >Still using Solvespace? It's lightweight and I want to do as much as possible in parametric design (CAD).
Open file (139.15 KB 1298x984 helmet-hair-V1a-001.png)
Open file (124.00 KB 1298x984 helmet-hair-V1a-002.png)
Open file (134.70 KB 1298x984 helmet-hair-V1a-003.png)
>>13710 Two other parts which I didn't even upload yet, are my first test on a 3D printable 'helmet hair' which would be printable in TPU and would then house some parts, so the rest of the head would therefore be smaller (if anyone wanted to complain about weight). This is also just some idea I want to try out since I saw something in an anime inspiring me to that. Doesn't mean I'll end up with that, but I'd like to develop it a bit to be an option to 'normal' hair. The helmet-hair could be printed with low infill and then be quite soft, but still house some electronics. The other one is my idea for a loaded-spring actuator for the arm, where the motor in the elbow loads a spring via some kind of chain, while the arm is down. Some mechanism, most likely a solenoid, would then block the spring till the arm needs to be lifted, and add some initial force to the whole thing. The chain or 'snake' would internally bend around the elbow and moving up while doing so. Don't know if this is going to work, but I intent to try it out. Some of the robots from therobotstudio use rubber bands to do something similar, which inspired me to try something like it. But I'd prefer a spring over a rubber band since it should last longer.
Open file (260.05 KB 1120x1132 1607917109192.jpg)
>>13702 Impressive, Iam far from doing research on skeleton mechanism with the help of CAD tool due to the lack of knowledge of that, I barely know how to make a bunny with a 3D printer lol When I will start doing the skeleton design, it might be a good idea to take a closer look of CAD mechanism design, thanks for the link.
Open file (461.10 KB 1024x613 SummerGlau-TSCC19.jpg)
>>13717 Solvespace and probably parametric design skills in general aren't that hard to learn. After going through some tutorials it's just about training and experience. Difficult is it, if you want a skeleton being very close to a human one and that might require combining CAD and sculpting or importing STL files. If you just want some tubes which move, that should be easy to do. I'm getting stuck on being to ambitious sometimes, then failing, getting a frustrated or tired of it and then getting distracted. Problem is, I'm going for the most ambitious version of robwaifus - the very human looking ones.
>>13720 Nice, I think Ive already heard those word years before, the parametric design stuff, thanks for the sum up. Ye, that's the problem with project, if we do not get fast result we can be happy with, we can be quickly demoralized. I wanted to make the perfect robot in the first design draft but the reality is the lack of money and I wish to build an acceptable robot sex doll at around 10k $ It's a real challenge and I cannot afford making very complex mechanism since it will cost too high at production chain. thus the challenge is to find existent mechanism and to deal with it. So I did it with this 3rd prototype and when I seen the leg moving up, it was so exciting heh, the first discover of a new lifeform :D
>>13720 https://www.aliexpress.com/item/33007068179.html Here, a 100cm humanoid that cost €35k lole I wonder if this robot can move with a silicon skin on
>>13753 Not sure if this posting was serious or if you're trolling. The 3rd prototype refers to the one in the first posting, or to the dog? >>13766 Too expensive and not good enough.
>>13786 lole, the 3rd prototype was in the first posting ye and this dog is here to fill the post space :D
>>13796 Then I'm glad, because of this concerned me: > when I seen the leg moving up, it was so exciting heh
Open file (92.61 KB 1080x816 Imagepipe_22.jpg)
>>13704 So, I'm on it. Time to make a separate thread soon, I guess. Tested the twisted sting actuator a few minutes ago a little bit. It felt strong, but my whole setup is currently too finicky to do propper testing. Getting the connectors and holders printed was also a bit harder than anticipated. The holes on the lower end of the motors aren't really parallel, I guess. Will need to improve that, now only two screws hold the "arm". Getting the twister on top (holder for the strings) right was also quite some trial and error. Also ran into another issue with my printer because I forget to fasten the nozzle while it was hot. So it created some hot plastic mess around the hotend, and the nozzle came out further through the pressure and so it got closer to the printbed nearly scratching it. Whatever, the robowaifus will come.
>>13931 Good to see progress. Do keep in mind that she will need a agonist and antagonist pair. I would recommend adding gears so that one pair winds as the other unwinds.
>>13937 >adding gears Ah, I think I know what you mean. I'll keep that idea in mind. Thanks. However, I plant to use very different muscles dependent on the movement, and this is a very early design. I'll try at first to get an arm lifting itself up and then going down through gravity after release. I also plan to add a cycloidal drive to the ellbow.
>>13962 You should draw a general picture of what you're trying to make. It will help your fellow anons understand your intentions and provide better guidance and advice.
>>13970 Maybe I will. Though, I haven't decided on every detail myself. It's more of a general thing to mention, that the arm won't need the same force to lift than to fall down again. Then for pushing something, another actuator might be better than a twisted string actuator.
Open file (122.64 KB 1080x810 Imagepipe_24.jpg)
I printed some parts of the spine prototype based on my design above >>12923. I printed one in full size and five at 40% or so, to build a little model without motors but screws instead to test it. Just need to buy some glue, and then design and print a rack for the whole thing. Also printed some shoulder part here >>9142 to test it. It helps to have it in my hands to imagine where to go with it. Now I have some more specific ideas how to use it. I already flattend the edge, so I can add motors to both sides. Other movements will probably be done with some string actuators.
>>13998 Glad you're learning how to improve it as you go. Do remember that walls bring prints strength, not infill. You can save a lot of mass without losing any strength by keeping parts like this 2 walls with 5 percent infill to help with bridging the large flat tops.
>>14003 Yeah, I was using 15% infill and two walls, which is probably more than necessary for a test. Thanks for the hint. However, printing isn't what's eating my time, designing and doing other unrelated things takes much more of it.
>>13541 OP, I'm going to be merging your prototype thread into our already-standing general prototype thread soon. Please keep us all up to date on your work, good luck.
>>14092 Oke and about this project I cannot do much now since the ongoing prototype will cost much more, thus I pause this project and do others meanwhile. Finding a job might be a good idea for the sake of that project lol
>>14092 Maybe it would fit better into the prototypes and failures thread. It's a prototype, not general engineering, and also conceptually closer to my Genera Plattform than other approaches, and I'm posting my files there for now.
Open file (99.44 KB 1080x629 Imagepipe_23.jpg)
>>13998 So, my progress is still very sluggish. I looked a bit into imageboards and hosting while the board seemed to be under peril. Got a bit stuck on 4chan as well, and also looked into which Linux distro to choose for the future. Which is kind of related anyways. However, I started to work on the rack to hold the spine. Picrel is just held up by a toothpick, and that's not what I meant. I'm designing a small holder with manual twisters to roll up some strings, so I can see how it behaves in this small model. Then I'll print one which holds the motors for the bigger one. I really need only a few hours per design, to get at least something interesting, so it should go faster now
>>14111 I started to use the references for the bigger one, till I realized this could be a problem if I want to use it on the small one. I probably can't just scale everything down in form of the STL file, bc some parts might be to small then. I hope I'll finish this tomorrow. - I will also need to buy some protective goggles to use while cutting down steel tube for the bones. - I was also already spending time on renaming the different workplane layers in my 3D files, so I can upload the files as polished version. It gets messy without proper naming. I already try to start naming these layers directly after creating them, so I don't have to look into it later and also to make my own work easier. - Then I also need to write some intro for the first posting to finally create the official thread to this project.
>>14111 Just some helpful advice, your printers bed needs to be leveled and you have some minor over extrusion. I've had those problems too and fixing them sooner is better then later. As for your design, it's pretty complex. I mean you'd just fine having a few with a semi flexible spinal material running through them. It would make the design easier to articulate while reducing mass. https://www.youtube.com/watch?v=EUEp-AfvvzE
Open file (108.20 KB 891x1564 media_ElTN4HJUcAIiR3U.jpg)
>>14113 Sorry, but I have to more or less deny all of this. My printer bed is quite well leveled out. I also don't believe you can tell that from looking at these pics. Then I know about the overextension, but I don't think this is important. Getting FDM prints perfect can also be quite a waste of time, and here its about prototypes and internal elements. I spend way too much time tuning my printer when I got it. The real problem I detected is, that gluing the parts together can produce irregularities. Might be less of a problem with the bigger parts, though. I also wasn't careful when I worked on that. As for you alternate design, I'm not interested in a simplified model. Otherwise I would have gone with that. It absolutely doesn't fit into my plans. I'll make it even much more complex, by adding a human-like s-curve to it later. This is the most simplified version already. Though, I might try out other approaches as well.
>>14102 >>14104 Alright, will do. I'll merge your thread here soon: >>418
>>14114 Hey, spine-guy, how much do you actually know about the human spine? I've been trying to figure out which thoracic vertebra would be the hardest to reach with the hands. I was thinking of having a rod sticking out of the back, somewhere it couldn't get in the way, to help support the body while trying to figure out movement. From this diagram I'm going to guess it's the 4th or 5th from the top.
>>14159 Sorry I don't know and I'm not sure if I understand your question. If you look at the back of someone bending forward, afk or in a video, you'll probably see all bones of the vertebrae. I think the ribs around the 7th from the top might be harder to see and feel, because of the shoulder plates and muscles.
Open file (93.43 KB 605x720 Imagepipe_25.jpg)
>>14112 Still on it, but it's trial and error. I made some mistakes and had to change and reprint it every time. Now it's glued together. Though, it hasn't been tested with the strings yet. The final version also needs to be printed with more infill, since the base gave in to pressure I had to put on it. I also took care of overextrusion like >>14113 above recommended. I had the Gcode for it in my config already anyways. The print still isn't pretty, but this doesn't matter and can be minimized easily. Then, I got a bit into face design in this paper craft style here >>271, but printed out of plastics, like mentioned here >>14124. It's also just a case study or early prototype for now, though. I'll post the images from my other computer.
>>14165 Your project has much potential. It'll be a long road to get things just right, stay strong. >Papercraft face Please do share your work in the waifu face dev thread. We can all use this. Face development is one of the hardest nuts to crack.
>>14165 I'm working with my current spine model now, trying to add some strings. Which is as annoying as it sounds. The thought crossed my mind, that this approach was a mistake and I should've printed it bigger from the start, since I'm using cheap plastics anyways. I'm glad that the twisters at least hold onto the knobs at the plate. I was concerned that I might need some more complex design and redo the whole thing. >>14167 >Please do share your work in the waifu face dev thread. We can all use this. My current policy or MO is to upload my prototype designs here in this thread. Then I add crosslinks to the other threads where it might fit in, which ideally other people should do more as well with their postings of designs. I also upload my designs to my Megashare account after a while. It's some additional work to do, so I only do that once a week or so. I then intent to link it again sometime in the threads covering some specific topic, when I reach a real milestone. I think the papercraft inspired chibi design in plastic idea is promising. For placeholders or simple robowaifus. The face can be a little bit 3D, and it could alternatively also be molded with silicone to be soft and skin-like. Designing and printing a fitting head should also be rather easy, in a simple form or more complex with a round back out of TPU (soft plastic). However, as you can see, I'm not an artist with drawing experience, so I'll need to do some work on it. This early design also has some error, indicated by the red edges. >Face development is one of the hardest nuts to crack. Yes, it's even harder if you want some kind of skull and soft skin. Everything very human-like. The Alita Battle Angel bust is the reference for the top notch model, but it also has to move, needs a mouth with a moving tongue and needs to have facial expressions >:-D ... long way to go :-c
>>14160 I really don't know how to explain what I said any better than I did. Trying to scratch your own back with your bare hands, which vertebrae do you think would be hardest to reach?
>>14170 I can reach all, but I'd say 6th or 7th. However, I don't really know.There might be articles to find on this.
>>14168 I caught some passion for that side-project, and went on designed a simple head. Finally found out about how to pick any color, since the color picker doesn't really work. So I can use some skin tones now, besides orange, yellow and brown. However, adding even more and more complex hair made my system starting to get sluggish, at least while having other programs open, so I might look into installing my Linux on the laptop first. Or I can still work on the spine and the face. We'll see.
Open file (61.59 KB 596x720 Imagepipe_26.jpg)
>>14168 I think I have to give the spine model a work over. I didn't consider that it wants to fall over to much in one direction. It's also difficult in general to roll the stings up and to keep it from moving in unwanted ways. This might be in part due irregularities in the balls I'm using, and in part simply because it is based on spheres in the first place. Now I'm considering something even more similar to the human spine, since this might work better: Something flatter instead of balls, but with round corners. Maybe also higher at the back than the front to mitigate the tendency to bend backwards. I considered such adjustments anyways, but I thought to introduce them later to keep it simple at first. I also had the idea of putting up a paper diagram on some door with all the parts I'm currently or soon working on, with a progress score attached to each. Like 'Arm TSA 1', 'Pelvis 0' or 'Knee 1'. Currently it has 15 points with low scores. From time to time I'm picking the one I'm most motivated to work on, instead of getting stuck with one part.
>>14199 I made some progress during the last few days, but nothing big. First, a kind of rig for my current shoulder joint design, with also some holes added. I want to try out different approaches in terms how to add motors or muscles to it, so I came up with this. The red holes go through the part and may be able to move the other part attached on the other side (right/outside) in two different directions. Obviously, a motor could be at the joint (in pink) moving it in both directions, but I want this to be fixed in that direction while strings pull the other part, which might be the arm, along the bigger part. In this test rig the outer part may or may not be fixed to the plate and then move the outer part or the big blue one. I also started do to design a "simple" elbow joint, which got more and more complex. It's just for testing my twisted string actuator, while not having a motor in the elbow for now. The red bars are supports. I added them in the design, so I don't need the slicer to add them, which would be much worse. Screws should go into the place where these bars are, though I might try to put in some roller bearings in a later version. Then, I had to come up with a new way of designing an eye for my face. Because the old one didn't work and trying to fix it cost me time without any gain beyond some experience. The new one is just a sketch so far, but it's way simpler to change. Surprisingly it's easier to use a Bezier spline with defined constraints, than using arches and line segments. I also made a calculation, that I could more or less design on part in one day or make a mayor improvement in the same time. Even if printing, testing and correcting takes another day or a bit more and I'll have sometimes other things to do, or slacking off for some time, it should be possible to create 100 parts or improvements per year. I plan to use this as a benchmark for myself from now on. Hundred new designs or improvements per year should be the minimum.
>>14274 Very impressive work Anon. Your efforts just continue to improve as is evident for all to see. On another level, you're also an encouragement to me personally. I too have been working to improve my craft towards achievements of reliable robowaifu software. It's a long and challenging road for me, but I too have made some nice progress in the last few days. I'll plan to post an concept intro release sometime over the coming weekend. Please stay encouraged Anon, you're a blessing to us all.
>>14199 >I didn't consider that it wants to fall over to much in one direction. Yes, gravity is a swine when it comes ball joints (well, any joint not linked to a powered servo, really). Can't really model it without a proper simulation either, so it's easy to make mistakes. I found it was best to constrain joints to certain axis of motion like X and Y. Which you appear to be already doing in >>14274, so nice progress! One question tho, anon. What are the threaded rods at the bottom of the torso going to be for? Are they intended to let the whole spine assembly move up and down?
>>14275 Thanks. Looking forward to it. >>14280 The problem with the ball joints is something I should have seen coming, but overlooked it. I'll find another way. I remembered, that my first ideas was to go with some kind of saddle joint anyways. Just went with the idea of ball joints on a whim. There are other options. >What are the threaded rods at the bottom of the torso going to be for? Are they intended to let the whole spine assembly move up and down? They are meant to hold the whole thing in place, till it's pulled in one direction but still restrained by the other stings. Which was a naive idea, I know. Wanted to try it out and thought it was easier to do so than it really was. These rods also represent the twisted string actuators I plan to put into the full scale model. >>14274 Today and yesterday night I worked on some improvements: I gave up on the bad design of the eyes in my test face, and implemented the new approach I mentioned before, based on Bezier splines. It's still not good looking but the underlying technique is way better now, and I can change it more easily. I'm glad I found that way by just trying out something. It only works with creating the Bezier spline in a certain way, otherwise it has more than four points which need to be constrained. Though, the whole idea of the printed face might go nowhere, since printed paper faces of anime characters might look better anyways. Then again, not anyone has a color printer for dead trees or a printshop close by and there might be other benefits. It's slightly 3D for example. The old design was severely flawed, because I made the mistake of picking the wrong layer to build upon. This is something to look out for, happens sometimes to me and can be very annoying. Let's say the eyes and eyebrows weren't really attached to the face in the old design, it only looked like it from the front, but there was a gap in between. That is solved now. For the future I will need to look at some more anime faces to see if and how I can improve the new design. Then I had to add some supports to my shoulder joint design, which I don't even know if it will go anywhere. I have a few other approaches in mind as well. However, the holes for the strings needed some support and I prefer to put such into the CAD design myself. This also proved more tricky than anticipated since it's fairly complex. I printed a sized down model of it and it looks good so far, but the test setup still isn't finished. Then, my elbow joint also made some progress and has been (partially) printed. The photos of my prints are on my tablet, so no photos here but nothing special to see anyways. I hope to focus on the elbow joint and motor holder next, so I can test part of my arm design with the twisted string actuator soon. Also need to cut the tubes and need to make the electronic wire setup less flimsy first, though.
Open file (35.12 KB 720x540 Imagepipe_27.jpg)
Open file (46.62 KB 683x697 Imagepipe_28.jpg)
>>14299 Today was a bit of a dud. Got a clog in my printer. Most likely kind of my fault, I hope, since this is better than a problem with the printer. Fell out of routine during my time without printing, and forgot stuff. I pulled on the filament to get it out, while the nozzle was at 100°C. Should have heated it up to 160-200 first. It broke within the heatpipe, and now I have to disassemble the thing and push it out with a toothpick (again). The upside is, that I'm getting more and more used to this sh...stuff. Like a soldier reassembling his weapon. However, maybe I'll still model something today, so I'll get something done. Days with archievements are just better. The pics are from yesterday. Got my waterbags, for testing my idea with using something alike as a muscle to hold the body bend in some way, without using the motors: Two bags with air, connected in the middle through a valve. Air is on one side through some movement, valve closes, body won't move back to the old state.
>>14312 I had difficulties the last time I tried to design a motor holder mentioned here >>13931 since the holes on the bottom are kind of skewed. I think I found a way of getting around that, by looking at it in a different way. It's important to ignore the openings for cooling, and just focus on the relation of the holes to each other. Also had an idea for the holder of the motor, attached to the arm. I will probably use some slider lanes. I'll put the screws onto the bottom of the motor, sticking out, then it can be hanged into these lanes. Only the heads and some plastic would hold the motor, though. However, I could make the screws longer and plastic thicker, or at some point later make the holder at least partially out of metal. Didn't print any of for now, since I didn't fix my printer yet. But I will, tonight.
>>14327 Is the pic #3 the one with 'slider lanes' Anon? Those look like they will hold another piece that fits into it very securely.
>>14327 These designs are very helpful for visualizing the general support structures inside a robowaifu's shell. Thank you for making these.
Open file (81.63 KB 720x520 Imagepipe_29.jpg)
>>14327 My printer works again. I'm really used to disassemble the head by now. It's still stressful, but I won't put it aside for two months again. After printing my elbow design, I found several things to improve, some are already implemented. This includes the issue that I didn't think all things through and I have to do it now. Especially how to exactly holding these part together. No big deal, some more modelling, with try and error. Currently it is way to noisy when moving them around, even without motor. I'll make sure to reduce the friction and let smooth metal run on smooth metal. However, a more general test with a more crude design and the twisted string actuator might come first, since this is what I'm longing for. >>14328 >Is the pic #3 the one with 'slider lanes' Anon? Yes, pic #3 in >>14327, pic #2 should give you the idea. The screws are like a cross, the two screws in the middle need to slide down further.
>>14299 >Thanks. Looking forward to it. Y/W. Here's the little intro project Anon. (>>14353)
>>14349 Several things kept me from maintaining my speed. First of all, I started to think my current approach through a bit more. I wanted to build an arm out of tubes first. However, even a 18 mm tube might be a bit much for the lower arm. I probably should go with something more like picrel 4 from Hamcat_mmq, but with a motor. I also realized that with my current approach the elbow might get too thick, especially since I want to fit in a motor into the elbow later. I realized that if I don't try to get the full tube size from one part of the arm to the other, then I only need something very similar to two shelf holders connected with a pivot. I actually already tried to build something like Hamcat did, though not very good (picrel 5) While thinking about those things, I got back to work at some point and tried to finish the motor holder (picrel 1-3). Well. I had and still have problems, because I don't know how to properly add skewed parts. I need it to be skewed in relation to the arm (picrel 3). Maybe I can do it with two different parts on the bottom, to connect the tube, which I then screw together. Though, the problem also affected my ability to design a hook (picrel 1-3), which I need to test the whole design. I don't even know if that is going to work. I had the idea to make a hook and hang it onto some knob of my drawer, so I can test the motor. Then I also ordered proper connectors for my ESC, since it needs XT-60 connectors and I only have the breadboard connectors to stick into it, which always fall out when the motor moves wildly around. Though, at least so nothing can go to wrong, since it stops automatically. Finally, my printer has problems again. I managed to get brittle PLA, and I think a part is stuck in the extruder. So I might need to disassemble it once again, or twice if the heat break is clogged again.
>>14373 Turned out my printer is fine. It's just that cheap (and a bit old) PLA filament might get brittle if it rests without being rolled up again: https://youtu.be/SvpSxHLotMI - don't have a Boden tube but direct drive, but this is what happened. Gladly, I was able to push out the broken filament without disassembly. I'm currently working a bit on that lower arm, since I looked at it again and think I can do it. Related: >>14397 I could do more, but I'm watching a bit too much YouTube again...
Open file (37.60 KB 1199x984 lower-arm-T1v1d-002.png)
Open file (74.37 KB 1199x984 lower-arm-T1v1d-005.png)
Open file (52.34 KB 1199x984 lower-arm-T1v1d-004.png)
Open file (43.64 KB 1199x984 lower-arm-T1v1d-006.png)
Open file (73.38 KB 1199x984 lower-arm-T1v1d-007.png)
>>14399 Looks good, but failed for now. I still think this approach to the lower arm is what I want to go with. However, after printing a miniature version I realized that it is harder to do than I hoped for. Ideal would be some simulation with movements, that marks or even subtracts conflicting parts of the design. The white bone needs to be able to role left and right, but in either case it needs to roll over the blue bone. In the best case, these two would touch each other the hole time, so they would support each other. However, since this might not be necessary, that's one ideal that I should better give up. It might have been the reason why it failed. I might be able to do it without such a simulation, but it will take quite some trial and error. So I should focus on other parts for now. I'll try it again at some point. Being able to print a miniature really helps. This whole thing is generally just part of the bigger ideal of her being very human-like. Making humans is hard. So, maybe I should go with something simpler for the start.
>>14474 Oh, I forgot: Picrel one above is something I gave up on already, but I might need to revive later. Cutting out the white bone from the side, using a Bezier spline. I messed up saving the file after I did that, though. Then I thought I don't need it anyways, which kept me going.
>>14474 wow, that looks like a very clever approach. nice one anon.
>>14474 I made progress with my elbow and arm design, but can't test it right now at nighttime. I already connected the motor mount and tested it, including the strength of the twisted string actuator a little bit. I had to change some parts of the "simple" elbow hinge, which took me some time. Also needed to create some mount to attach the whole thing to my drawer, so I can test the arm (picrel 1). I also started to work on a design for a pelvis (picrel 2 and 3). I planed to use a stl file based on a human bone model, slice it, simplify it and make it malleable with parts in parametric design. But now I started a model from the scratch. Don't know if that will work out, but I can do it now and it might be less stressful. It's just important anyways, that the parts which are visible or perceptible to touch through the skin are similar to a human female. Internally I need to use the space differently, since for example the hips should hold two motors with cycloidal drives.
>>14567 > It's just important anyways, that the parts which are visible or perceptible to touch through the skin are similar to a human female. This. Really good insights, Anon. There's little doubt that doing our best to mimic real female body structures will greatly enhance the pleasure and comfort of owning such a robowaifu.
Open file (24.38 KB 255x720 Imagepipe_30.jpg)
>>14567 So, technically the first build of the Genera Plattform moved a limb for the first time today. Some minutes ago. I got the elbow moving, also with various provisional arms attached. Of course, I directly ran into several problems, but it still looks very promising. - The motor is not a servo. It can't measure where it is, nor can the arm signal it to the motor where it is, since there are no sensors at all yet. So, the arm goes up, then I have to turn the motor of and the arm goes down. Not a problem. Won't be the only actuator anyways and I realized that it only needs to lift the arm up to 90°, then a weaker but more precise one can take over. I also plan to track the position of the arm of course, and hope the twisted string actuator will be a bit more controllable that way. With no load, I can even not look at it, only give a brief burst and it is up. - Brings us to the other problem, that my holder for the motor isn't very strong, so if the movement goes to far then it takes the whole thing apart and even twists the cables of the motor. - Last but not least, rewinding the strings is a problem. Though, this might be in part because my test controller for the bldc motor doesn't do reverse. Just adding a counter actuator will definitely not work. The strings can twist so hard that it is difficult to unroll them. The arm going down also doesn't mean that it unrolls. Maybe then sting is to flexible and long. Need to learn more about it.
>>14572 Little clarification: The arm might need to go up 90-180° till another (mybe weaker) motor can take over alone. It depends where it starts and if the shoulder moves as well.
>>14572 >So, technically the first build of the Genera Plattform moved a limb for the first time today. Grats! You're making solid progress now, Anon. I also appreciate the improvisational approach for adapting to the environment for you robowaifu R&D lab, whatever the circumstances! :^)
>>14572 Great job but, your motor mount is very thin for an actuator with the kind of power that bldc can give off. The base would also benefit from being able to pivot. You will need to implement a controller that can reverse and an opposing twisted string so the arm can move forward and back as needed. I also recommend using a non backdrivable geartrain to prevent unwanted unwinding.
>>14592 >motor mount is very thin Yes, I'm going to make it stronger. Currently the problem is rather that it is even only sticked together. Not one print nor welded or glued. PLA is very strong, so I'm not worried about the strength of the part, really. >The base would also benefit from being able to pivot. I don't know how you mean that, but I'm going to watch some TWSA videos, which might have similar advice. >need ... controller that can reverse That I already realized. >and an opposing twisted string That you told me before, but I don't know how yet and not sure if I need it (at this point). The arm would fall down due gravity, just by unrolling the string. It even does so, by stopping the motor, but it doesn't unroll completely. >also recommend using a non backdrivable geartrain to prevent unwanted unwinding. Thanks, I'll look into that.
Open file (124.36 KB 533x502 Capture.PNG)
>>14594 Hi, long time no see I have an idea that aim to resolve the freedom of rotation of joint. It's to use a stainless steel chain and to allow roll axis maybe adding more links will make the roll rotation even better. I am skipping the joint ball prototyping because it's too much difficult to put it in place. Hope I will be able to rework on that project soon, it miss me so much heh. Cheers
>>15293 Hi Evodoll, good to hear from you!
>>15293 Welcome back Evodoll, your idea to use a chain a flexible shaft is pretty clever and I look forward to seeing how it works out. Wishing you the best!
Open file (1.41 MB 890x892 1645696190272.png)
>>15303 >>15304 Thank you guys, challenges await me !
Open file (2.58 MB 4128x3096 20220321_171645.jpg)
Almost done developing a simple easy to print and assemble omni wheel. Most of the ones I found online for printing are not designed to be circular and are harder to assemble. This just requires bar-b-q stiks which are eaily gotten at the Dollar Tree rather then screws. This is the first working protoype with a 4 inch (100 mm) diameter.
>>15686 Brilliant work lad. So question; is this primarily to be a passive 'roller', or do you intend it to be an active 'locomotor' eventually? >This just requires bar-b-q stiks which are eaily gotten at the Dollar Tree rather then screws. I assume some kind of glue or other fastening mechanism is req'd for the stick ends? Or maybe they're tight fit enough? BTW, I greatly applaud your focus on 'cheap & easy' Anon. This is exactly how we'll see robowaifus reach a very broad segment of men everywhere.
Open file (950.59 KB 2827x2393 20220322_083327.jpg)
Open file (921.21 KB 2986x2623 20220322_083348.jpg)
>>15689 Designed to work both passively and actively. >Maybe they're tight fit enough? This is correct, designed to be easily put together by shoving the stick in then cutting the end off. Here's the next result, it works but is a tad big at 2.5 inches thick and 4 inches in diameter though it barely costs anything to make.
>>15695 Curious how the locomotion version is going to, well, locomote Anon? Do you have a clear vision worked out for that yet, prototypes maybe even? I'm really glad to hear the manufacture is both easy and inexpensive Kiwi. Perfect combo! Cheers.
>>15761 By rolling Chobitsu Kun. The friction between wood and PETG/PLA is pretty low. Right now I'm trying to make the mechanism flatter and print faster. PETG/PLA also don't have a high coefficient of friction on most materials. Due to this, I'm looking into materials that will help in the transfer of kinetic energy from wheel to surface. I'm thinking of O-rings at the moment.
>>15765 >By rolling Chobitsu Kun. haha, make sense. I suppose your mention of the O-rings, along with your previous posts about elastic bands as force & rotation drivers pretty much answers my questions. It's going to be great to see unified, modular roller 'feet' available to all our robowaifu kits Kiwi. Drive on! :^)
Open file (6.32 MB 4624x3472 IMG_20220416_125509.jpg)
Nylon planetary gearset with 78 ring teeth, 32 planetary teeth, and 12 sun for a 7.5 gear reduction ratio. I fucked up the ratio (78-12)/2=33, which is the correct number of planet gear teeth. I also had bad warping on the gear teeth.
>>15890 Looks really cool. But if the nylon is warping why not try printing it in PLA?
>>15890 Looks pretty cool though Anon. Can someone here explain to the uninitiate what a planetary gearset is for? I mean what is the primary purpose/benefit it serves? Can they improve some things?
Open file (6.06 MB 2851x1997 RyoZU.png)
>>15894 >What is a planetary gearset (they are also known as epicyclic gearing)? A gear train which relies on three basic element types: sun gear in the center, planet gears which "orbit" the sun gear, and a ring or annular gear which has internal teeth that engage the planet gears. The calculations for gear ratios depend on what element is being driven and which element is used as the output. This website provides a very helpful visual interface for understanding the relationship between the elements: https://www.thecatalystis.com/gears/ This website provides some good general knowledge with helpful explanations on the math involved in calculating gear ratios: https://woodgears.ca/gear/planetary.html >Why use it? Planetary geartrains are legendary in many engineering applications for their high efficiency (generally over 95%) and their ability to create large gear ratios in compact spaces with few gears. A notable advantage they have for 3D printing is the distribution of torque across multiple planets which engage the sun and outer ring. This allows for a higher torque density (the gears can be smaller than other geartrain types for the same torque requirements). Though I would suggest always designing 3D printed gears to be thicker with stronger teeth than is required. They also look cool. (Pic actually related, she is a clock/automata with planetary and many other geartrains)
>>15899 Ahh. So, could the outer, ring-gear be affixed to say, a thigh-strut framework, and the inner, sun-gear be affixed to a drive motor? Sort of as a way to increase the torque for driving the upper leg motions of a robowaifu? Is this a legit way to understand these things or no? >(Pic actually related, she is a clock/automata with planetary and many other geartrains) She's pretty wonderful robowaifu too.
>>15904 >Outer ring as output, driven by the sun gear, with stationary planets. That's definitely an option, would look really cool too. It's really nice to find someone else who appreciates RyuZU, she's been a favorite of mine since the Clockwork Planet anime aired years ago, actually got me into automata and clockwork mechanics. Hidden for off topic
>>15907 Great thanks! That helps me out understanding things quite a bit Anon. >also >inb4 Kiwi is henceforth known simply as the entity 'Y' AnchoR is breddy wonderful too. In fact, scene-related is one of the most touching moments in all animu IMO. Certainly it touches on an incredibly broad swath of topics and interests for us here on /robowaifu/. >
>>15909 Finally starting get the hang of resin printing. It's actually pretty easy, and with way better mechanical accuracy then filament prints. Shout out to RiCODev, his advice saved me a lot of trouble and trial and error. No idea what he was going on about with needing supports, I don't use them and everything is fine. Don't overshake your resin, your prints will have many bubbles like these do. Also, you must be very careful with cleaning the build plate after every print or else your print may end up stuck to the platic on the bottom of the resin vat. The key to successful designs is to shrink the part which touches the build plate to accomodate the first few layers expansion. Also, larger prints should be hollow with holes punched into the shell to releive hydrolic pressure, very grateful to have learned that from RiCODev's warnings. Also, rounding edges to create continuous edges helps, as you can see the overhang is a continuous rounded feature. >Y Very tempting, I've been thinking of changing my name as I do not wish to be associated with KiwiCo and other existing businesses that I'm not affiliated with.
>>15904 You'd usually get your highest gear ratio from having the ring embedded in the robot frame and the output shaft connected to the planetary gears, with an "carrier arm" attached to the bearings.
>>15914 I hate my ender printer and im sure sticking to the bed in the first place would not be as much of an issue with resin printing. What would be a good resin printer to buy?
>>15926 I've only used the Anycubic Photon Mono 4K so, I would have to recommend that. This is a good price for it: https://www.amazon.com/ANYCUBIC-Photon-Resin-Printer-Clear/dp/B09Q93Y374 I too hate my Ender 3, though for bed adhesion resin printers need the metal plate to be cleaned between every print, I use a paper towel to do this. I would also recommend a magnetic plate for the bed like: https://www.amazon.com/Sovol-3D-Upgraded-Platform-Magnetic/dp/B08LYP2VTT/ref=sr_1_2?crid=155V9SBK8WNCS&keywords=photon%2Bmono%2B4k%2Bflex&qid=1650305462&s=industrial&sprefix=photon%2Bmono%2B4k%2Bfle%2Cindustrial%2C184&sr=1-2&th=1 these make print removal much easier. Wishing you luck and success in your printing endevours.
Working on connectors that will work with both resin and filament printers, does anyone know why my resin sometimes creates a large protrusion where a substantial curve meets the build plate? I think it may be my resin, it's Sunlu and seems oddly difficult to work with compared to my friends Anycubic resin. (This failed design can only sustain a few pounds of force before catastrophic failure, oddly always on the left side.)
>>15964 Good shots, Kywy. Nice detail, good focus. May I suggest you add a soft fill light (checkout '3-point lighting') as well? It might give us all a better look at the details on the 'backside'. Just a 'reverse' white placard should be sufficient for a shot lit as in your picrel. No need for a bulb. I'm unable to give you any information on the resin printing issue haven't done any myself yet. However, I want to encourage you again in your goals of inexpensive, easily-mastered construction techniques (and kits eventually?) This will be a tremendous benefit to the unskilled man who wants to create his own robowaifus but doesn't have any clue how to begin at first. >tl;dr Just imagine what the benefits would be to us if we could somehow travel back in time, and deliver our completed designs to our younger selves! :^) Really looking forward to your designs to come Anon. Indeed for all of ours. Onward! >=== -add 'white placard' comment
Edited last time by Chobitsu on 04/22/2022 (Fri) 04:52:13.
>>15926 I would encourage you to look at the larger machines (Elegoo saturn/Anycubic photon x), I know they cost a little more but the extra build area makes the difference between a toy with niche uses and a truly usable machine. Even if you only plan to do small parts, being able to shoot off an entire set at once is a game changer. I was able to make RiCOs arm overnight whereas it would have taken a few days on my old machine. The extra resolution on the small 4K machines only offers a marginal quality improvement. Not worth trading for build volume.
>>15974 Been considering a Elegoo saturn myself as it has a similar build volume to what I already have with traditional 3D printing. Is it really worth getting? Have seen a lot of reviews claiming their units were defective or talking about how it's hit or miss whether resin printing is hazardous material cleaning time or 3D printing fun.
>>14572 Brief update. I'm learning OpenSCAD now. I want to be able to change files with only a few commands.
>>15975 If you need smooth organic features (no layer lines!) then yes. Otherwise, no. See my post in the Rico thread. Yes, the resin is toxic until cured. Read the MSDS for common resins, TL;DR: wear gloves, don't huff it, and you should be OK. Cleaning IS a PITA, you will probably find yourself avoiding resin printing in favor of FDM whenever possible. You will need a dedicated cleaning area.
>>15975 In the 3D printing thread >>94 other smoothing methods have been mentioned. Resin printing might be only worth if you want to print small models which need to be smooth, or for ones which need to be precise and smooth. FDM is the to-go method for a reason. I would rather buy small metal gears from Ali Express instead of printing them out of resin, but for prototyping it might be nice to print some placeholders.
>>15976 Great! Glad to hear from you again, GenerAnon. Good luck with your new efforts, keep going.
Open file (19.55 KB 1025x673 chest-cut-v2.3-002.png)
Open file (77.59 KB 1025x673 chest-cut-v2.3-001.png)
Open file (23.60 KB 1025x673 chest-cut-v02-001.png)
Open file (33.83 KB 1153x715 thigh-test-01-nohull.png)
>>15976 Using OpenSCAD for this seems to be quite unexplored, but I had a hunch it would be the right way to go. Some things work quite simple. I wanted to cut out a chest from a cylinder using four half rings, and now I see even quite good results using just one. Other parts are just distorted spheres which are then surrounded by a hull, using the hull() operator. robowaifu-OpenSCAD-tests.tar.gz: https://files.catbox.moe/v8rdf0.gz
>>15993 Neat! Looking forward to your prototype explorations Anon.
>>15994 Thanks, I'm still at it and won't drop out permanently. Just having some phases of being demotivated or distracted from time to time, or having other things to deal with. Sometimes all of it in a row. Whatever, I'm back at it. Currently, trying to work out how I can make different kind of faces in OpenScad. My goal is to have a bunch of files where one is the full model, but changing a part will make the rest adapt. Within a certain range it should be possible to change every part along certain tastes. The internal parts should then also adapt to the outer shell. It's still gonna take I while to get there, though.
>>16680 Working mainly on the head and eyes for a while now, but also the lower legs. I'm trying to make a anime doll-like head first. However, I plan at some point to try for a more human-like version but still waifu (Alita-style) as well. Currently I'm playing a bit with the size of the eye cavities. If I can have different layers to it, making it more human-alike, or not. The current prototype face has no nose yet and it looks too much like a genderless toddler, lol. I'll try to shrink it horizontally soon, but currently the values aren't dependent enough on each other, so every change takes some effort. My earlier efforts looked a bit more like some furry-face. Probably because I looked at some figurine like Ren at some point. I changed it a lot since then, but I'll keep the old versions. I'm always making a new file with a new name from time to time. So if I go into a wrong direction or want to make another version of it, I can go back.
>>16680 >>16691 You're making progress Anon, so just don't quit. The fact you're doing your work in a CAD program actually puts you ahead of most others ATM. Keep it up! One of the well-established, so-called '12 Principles of Animation' is Appeal. Frank & Ollie had this to say on the matter: >#12 APPEAL >"A live performer has charisma. An animated character has appeal. Appealing animation does not mean just being cute and cuddly. All characters have to have appeal whether they are heroic, villainous, comic or cute. Appeal, as you will use it, includes an easy to read design, clear drawing, and personality development that will capture and involve the audience's interest. Early cartoons were basically a series of gags strung together on a main theme. Over the years, the artists have learned that to produce a feature there was a need for story continuity, character development and a higher quality of artwork throughout the entire production. Like all forms of story telling, the feature has to appeal to the mind as well as to the eye." https://frankandollie.com/PhysicalAnimation.html I would recommend you dabble with sketching human faces & animal 'faces' to get a general feel for the topic Anon. You're establishing some of the basic facial elements well; now chisel out the refinements needed to bring together an appealing character visage overall. Just keep.moving.forward., Anon! Cheers. :^) >=== -minor fmt, prose edit
Edited last time by Chobitsu on 06/15/2022 (Wed) 07:29:15.
>>16697 Thanks for you kind words. My goal is a kind of framework, which allows for putting out personalized faces and other parts, dependent on the personal taste of the user. Later we can hopefully marry that with some machine learning and then go from drawing or pictures to a 3D result. I discovered the website https://www.artbreeder.com/browse recently, which could be useful in that regard, or we need something similar down the line.
>>16702 Neat! Thanks anon. >My goal is a kind of framework, which allows for putting out personalized faces and other parts, dependent on the personal taste of the user. I kind of have something similar in mind for RW Shell (and dependent libraries). My current plan is to generate meshes from scratch using parameters fed into the library, and they can then go back and forth with Blender, OpenSCAD, etc. The desire is that this same content will serve both for the data for the practical 3D prints, as well as the Visual Waifu avatar models. Keep up your efforts, Anon!
>>16691 Currently working on printed bearings with steal balls (picrel 01) based on this video: https://www.youtube.com/watch?v=5YIZaJnER_k and also this explanation about using it for the base of a robot arm: https://www.youtube.com/watch?v=EsnlUA6U4kI Same time I'm printing the cycloidal drive by Matt Moses: https://alogs.space/robowaifu/res/12810.html#q16720 - Just for testing right now. I hope I can use it. I need a model that I can change well, also adding bearings to make it run smoother, additional stages or metal pins however I want it. I also tried to design a eccentrical cycloidal gear, but without knowing the right math, just experimenting around (picrel 02).
Open file (59.11 KB 1244x620 foot-005.png)
Open file (90.24 KB 1244x620 full_body-07-001.png)
Open file (51.58 KB 1244x620 full_body-07-002.png)
>>16735 I realized that the RollerBB of RoTechnic is more sophisticated than it looked at first. I thought about using his model instead of making my own. Tried it, but it didn't print well when sized down. It's really for replacing big bearings, but I want a model that works for smaller bearings as well so I'm going to work on it. Then I printed some cycloidal drive parts (https://www.thingiverse.com/thing:551243), with mixed results. I think it might work to use that model, but I need to make it more malleable to my needs. Then I also worked on my body model, with some progress and trying out some modifications, but nothing huge. Lower leg is there now and I'm making progress with the feet (picrels).
>>16751 I started to work on something that might be considered to be a bit OT. It's a holder to make the joints of my future "prototype dolls" moveable. This could of course also be used to build them as non-robotic lovebots. I consider the line to be fluid. That aside, I think it would be good to have the men covered, which only need something more simple or can't afford more. However, I used my cycloidal disk as a pattern, and try to model a mechanism where one can push a kind of wedge from one side, maybe putting the finger a bit into a hole (some knob would stick out on the other side), then move the mechanism around and therefore the attached joint like e.g. a leg, and then let the mechanism snap back in. On the back of the wedge would be one or several springs. The blue circle is the cutout for one. The blue cylinder in the second picrel is probably bigger than the mechanism will be. It's most likely not going to be visible (and of course not in that color). Except someone is open for a more robot-like look, or even drawn to it. I'm of course still committed to the more ambitious goal of building a robowaifu, not only dolls which need their joints moved by hand. It's just a "while we're at it"-side-project. Also it might be useful for the prototypes and to have a more gradual way for the future clients to get into the whole thing. The next level above such a doll should use cheap stepper motors, to make the body automatically poseable, but they would be rather heavy for their size. Still good enough for many use cases. I also made some small progress on other parts.
>>16735 >>16751 >>16772 Proud of you Anon, keep going!
Open file (69.41 KB 1080x740 Imagepipe_45.jpg)
>>16772 Had some printer issues, but I printed the first prototype. Clearances were too small, I need to shrink some parts to make it fit better. These towers or spikes going into the cycloidal pyramid should be screws or bolts anyways. Also, the button to press and the hole needs to be bigger, it's smaller than my pinky right now, but it should be reasonable easy to press. However, I can already tell that the mechanism works as anticipated. I can switch between 10 angles. Added to a leg it might be less than that, since there are obviously constraints of movement. Maybe I'll make a version that takes that more into consideration. I also want add a springiness option. If the button isn't pressed, the joint could still be moved by a human, but it would snap back with a low amount of force.
>>16803 Working on the torso right now, but it's rather difficult. Generally I need to clean a lot of code, making a lot of parts scale together and in the right relation to each other. I will likely go on working on feet instead, so I can print a sized down model of the legs with the feet underneath and some simple connection between the hips. I also made a version of the face with the eyes closer to each other. I take back the term "genderless" which I used before. The chin makes the face clearly female. Also, I realized the eyes are very far away from each other in cute anime characters, like Hoshino Yumemi (Planetarian). So I really need different versions and playing around with other parts. I also want to make a more human-like version soon, but still with bigger eyes.
>>16805 Glad to see you continue to make solid progress. May we bless the world with waifu's in our own ways. Really rooting for your success!
Open file (59.09 KB 1146x660 full_body-09-003.png)
Open file (88.38 KB 1146x660 full_body-09-001.png)
Open file (82.31 KB 1146x660 full_body-09-002.png)
Open file (36.51 KB 337x475 black-magic-m-66-1.jpg)
>>16805 Worked on the chest. Still looks weird, but I think I'm getting somewhere. For the very human look I need better integration of the existing parts, for a more artificial looking version it might not even matter. With the colors in picrel two and three it reminds me of M-66 (aka Sophie). Maybe I can draw some inspiration from there. I was cleaning up the code a little bit otherwise, trying to make parts like the chest scalable compared to other parts. So more variables need to be dependent on each other. Didn't do as much as I could've done, but I'm still on it. Listening to interesting podcasts works quite well to keep me glued to my seat for some hours. >>16808 Thanks, I'm glad about your progress as well. We need to get something to show for soon. It's about time.
>>16823 Made some more progress, but also have some funny fails to post. The chest looks a much better now, imo I really crossed a border there. Getting better at the merging of spheres. I also improved my cycloidal joint connector, so I can increase the size of the button for moving it. Made some other parts smaller, with bigger clearances, since the parts I printed before didn't fit together. I still need to check out the new design. Also cleaning up some code, which is not spectacular or visible, but important. I will upload some more files soon.
Open file (2.00 MB 1920x2560 PlanarDemonstrator.jpg)
>>16827 It's impressive how far you've come with Solvepsace. Have you considered using Blender for the face though? Organic parts are easier to model in that program. Modeling concepts in mostly flat fast to print parts is a great way to iterate on concepts fast.
Open file (61.87 KB 1244x620 foot-005-001.png)
Open file (72.89 KB 1244x620 foot-005-003.png)
>>16830 Thanks, but this wouldn't be possible in Solvespace. I'm using OpenSCAD. My older parts are in Solvespace, and they might be usable for that at some point. However, I might move them over to OpenSCAD later. Blender does integrate Solvespace now, which makes it more tempting. But the thing about these part is, that they are all in code, and can be changed by altering some variables. You can't do that with sculpting. I don't want to sculpt one face and then the skull based on it, and the next person has to sculpt another face and change the whole head including the skull. I ideally want autogeneration of 3d models of the whole head from pictures (like e.g. from Artbreeder.com) done by a deep learning network. Including the automatic generation of all the bones in the skull based on the changes, and then also the same for the whole body. On top of that, all the forms for molding the parts in silicone rubber. So the body parts can be printed, or molded, or printed with an offset to the molds and then have the silicone skin added to the printed parts that way.
Open file (48.13 KB 324x623 full_body-13-001.png)
Open file (55.12 KB 277x623 full_body-12-002.png)
>>16831 File drop, 42MB compressed to 12MB: https://files.catbox.moe/e7i363.gz Still messy and all over the place. Made some additional updates on the body. The butt is bigger than I intended it to be, it just turned out like that for now. It's because of the current height of the hips. I was mainly working on cleaning up the code recently, but also added new lines. Don't want to finetune the body type right now, looks quite fine anyways.
Open file (38.38 KB 327x623 full_body-14-004.png)
Open file (33.06 KB 327x623 full_body-14-003.png)
Open file (47.54 KB 327x623 full_body-14-002.png)
Open file (42.29 KB 327x623 full_body-14-001.png)
Open file (36.02 KB 327x623 full_body-13-006.png)
>>16836 I slacked of yesterday and went down another rabbit hole. But I'm back. didn't do that much so far, but something: I added the tiddies yesterday or the day before (don't remember). Today: - neck - back and sides - sacrum area The todo list is getting shorter: - nose - ears - arms - finishing the feet, especially toes - hands - better anime-like face - more human-like face variant - shoulder plates (spikes) - optimizations - code cleanup - body types via easy variables and changes - private parts (if not meant to be bought) This is probably still two weak of work, I guess. This is of course only the outer shell, after that the parts need to be separated to be printed and later molded. Printable molds need to be created. Also, the work on the internal parts need to go on. Creating a good TPU pattern for a pleasant and semi-realistic deformation of the body will also be important. Oh, I changed the color of the body in the model. Got a bit of an meat overdose. The more human she looks, the more I prefer her to look like she's wearing a body suit. Also keep some parts colored to show what changed, and it looks a bit more like a suit.
Open file (104.61 KB 1080x865 Imagepipe_46.jpg)
>>16847 My plan is to use TPU with empty space inside the shell. So it would be squishy. I still need to workout a better way to make it more squishy in different directions at the same time. Currently the part is stiffer along it's elongation, while the circles make it squishy when pressed together. I want better control over this, but I don't think this can be done reasonably within Solvespace, I need to do this in OpenSCAD. The files above are made in Solvespace. It's just a simple test so far. The upper part is the most advanced one. The pattern outside is meant to feel like muscle fibers even if there's some additional material on top, like fabric and silicone. Prototypes and cheaper models might use mainly such printed parts, while more advanced models might use more silicone rubber. The tubes inside could hold additional textiles rolled together to make it harder to squish, or these spaces could be filled with cheaper silicone than the one being used for the skin. The empty space could as well hold air muscles, and tubes for water, other liquids and air. Of course, also cables and sensors.
>>16849 Interesting. Remember you can print molds for silicone and use different stiffnesses of silicone for any desired effect.
>>16852 Yes, I hope to be able to combine both. TPU with hollow spaces should cut down on the amount of silicone rubber needed. Also, I don't want to make very complex molds if I can print it.
Open file (32.53 KB 425x623 foot_T02v01c-002.png)
Open file (58.05 KB 425x623 foot_T02v01c-001.png)
Open file (44.67 KB 425x652 foot_T02v01i.png)
>>16849 I'm still on it, btw. It just turned out to be harder and more time consuming than I thought. The time estimate of three weeks from before isn't realistic, it maybe still rather four or more weeks till I have at least the full body outer shell model. I'm still working on it every day for a few hours, sometimes more sometimes less. I would have even more time, but I'm still drawn to watch some Youtube and also have some anime to watch (and more important things I should do). I redesigned the foot from the scratch (based on the same pattern of female footprints) and working on the details. My Nema 17 motors from China stepped into some problems and will not arrive anytime soon. They were already in the country I live, but there were still some problems with documents and they were send back to China. So I have no real urgency to work on the cycloidal reducer. I'll just start printing simple models soon, mostly in the size of small dolls, just to look on the parts how they look printed and to test the printability in the first place. >>16830 >Have you considered using Blender for the face though? I forgot to mention in >>16831 that my whole approach can most likely be combined with existing models from other programs. At least STL exports, which can be imported into OpenSCAD. So I could take any part from someone else and use it. Though the problem in that case is of course, if I want to do more than a doll but go for animation then I would need to design that internal model for a part which I can't change itself in OpenSCAD. I might even try that at some point, but ideally we would have one system where the parts are completely related to each other. It might generally be more difficult using a STL file while doing the internals in OpenSCAC, since I could remove parts from a file of the face in OpenSCAD and so to speak "look into it" while working on the internal mechanisms.
Open file (19.78 KB 425x612 finger_test_001.png)
>>16855 I got a bit slowed down for various reasons, but I started working on the hands.Picrel is supposed to become a finger and then a hand. I'm not sure yet which will be the best approach, so it's entirely possible that I'll have to restart later, but it looks promising.
>>16872 >it looks promising. Yes, I agree. You're tackling the 2nd-hardest part of exterior human anatomy when you deal with our hands Anon. Godspeed.
Open file (104.29 KB 425x612 finger_T01v01b-003.png)
Open file (88.12 KB 425x612 finger_T01v01b-002.png)
Open file (75.52 KB 425x612 finger_T01v01b-001.png)
>>16872 I thought again a little bit about how well my approach is suited to the problem. I'm always coming to the conclusion that I would at least need a model of the shell for creating the internal structures anyways. Maybe I'll use another program for some parts later, based on my existing files. But then, those existing files in OpenSCAD format will help me to know where internal structures like bones or space for sensors need to be. For adding certain internal parts it might also be better to print some designs or mold them piece by piece, add the bones, cables and sensors, before molding the outer skin. I'm making generally good progress, as soon as I get myself to start with the work. As soon... that's sometimes the hard part. I found that listening to podcasts which are not to difficult to follow or have aggravating content helps the best to keep me going.
>>16893 You're doing a good job with the modeling of the fingers by the appearance of the shapes Anon.
Open file (60.30 KB 425x612 hand_T01v01d-002.png)
Open file (47.79 KB 425x612 hand_T01v01d-004.png)
Open file (37.14 KB 425x612 hand_T01v01d-001.png)
Open file (40.09 KB 425x612 hand_T01v01c-002.png)
>>16893 Only a little bit improvement, once again. Fingers got scaled to be more different. Backside is rounder now. Working on the palm of the hand. Thumb is still missing. >>16897 Thanks. Yeah, sometimes progress is fast, sometimes it's slow and tedious.
Open file (62.82 KB 425x612 hand_T01v01f-001.png)
Open file (117.27 KB 425x612 hand_T01v01f-002.png)
>>16907 Next little hand upgrade. Adding the thumb wasn't easy, but easier than I thought. I think I should move on to the next part and optimize later.
>>16907 >>16945 >sometimes progress is fast, sometimes it's slow and tedious. Ehh, you're doing fine Anon. Add the thumb really helps define things well!
>>16945 Cleaning up some code today and working on the option to cut the main body into pieces fit onto a printer (picrel one). Also for the next step, of designing more of the internal parts. I also added the option of creating a kind of measure stick on the side of the body to show the height of some part. Yesterday I had some more funny error, by changing one little thing, and ended up with some kind of strange skeleton (other picrels).
>>16949 Slicing down the full body into separate element is going forward. Same for making the resulting parts hollow. However, the hollowed part makes my PC sluggish when moving it around to look at it after the alteration, so it's on the border of what can be done comfortably with OpenSCAD. The long term goal is to have a Python program that creates the files of the parts from the main file. So one could change the full body, then run the external script which would create the files for different body parts. Still a lot of work to do.
>>16958 Your progress is impressive! It's pretty amazing how far you've pushed OpenSCAD
>>16946 >>16960 Thanks for the encouragement. In a way it's just going as suspected, though a bit slower overall than it could, since I'm not always working as much as possible on it. In that regard, I found working on anything where I can look at something looking like a female body is really very motivating. Big difference to working on designing mechanical parts. The great thing is, that now if I want to change something I can easily do that. Think of the angle of the bones or some connector between limbs, or the overall body shape. I thought modelling it all in a program where I have to click and drag to do changes would be a lot of work for every change and every test or optimization. It will be much easier and faster doing it in code, where I can move things around by changing some variable and also make things dependent on each other or relative to each other. >>16958 I thought about creating some sized down figurines for testing, but I'm not sure how yet. The whole approach wasn't meant for having a ball joint structure inside. My cycloidal drive isn't ready and the motors are still in China or in the sea. Also, the body is meant to be soft, not really hard shell. Though, I might try out both. Maybe the test figurine will be like a plush toy in TPU without PLA parts inside. I also have the cycloidal holder I designed for something like a poseable lovedoll. We'll see. I also didn't optimize the body itself yet. The crotch area is still too low, I think the whole hip area is to long in north-south direction, and many details all over the body are missing or haven't been worked out completely.
>>16962 I like your hip-socket design approach Anon. Should be quite robust in the face of lots of torques there. >In a way it's just going as suspected, though a bit slower overall than it could, since I'm not always working as much as possible on it. In that regard, Ehh, always double your initial estimates on timeframes needed then double that :^)
>>16968 >always double your initial estimates on timeframes needed Yeah, since you wrote that I had some motivational problems due to heat and related sleeping issues, other things to do, then a hardware problem with my PC, which gladly got resolved without ordering new parts. I'm back on it. Had to think a bit about how to proceed with one challenge anyways.
Open file (1.16 MB 1920x2560 ParadoxDrive.jpg)
First successful Paradox Drive prototype. A brand new gear system that functions just harmonic drives but, without a belt. Feel free to do whatever you'd like aside from patenting it. https://files.catbox.moe/byqa2g.stl https://files.catbox.moe/3vjb21.stl https://files.catbox.moe/dmrv61.stl https://files.catbox.moe/zw1w37.stl https://files.catbox.moe/cmwo9w.stl
Open file (491.22 KB 1192x894 gambatte.jpg)
>>17086 >First successful Paradox Drive prototype. Sweet! Remarkable insight, Kiwi. >1.21Gigawatts/10, would drive it again.
>>17086 BTW, I think you'll need to beef up that lateral transfer bar across the center Anon. Shear forces are likely to damage under heavy loads, I think. Cheers.
>>17089 It's made for demonstration, actual versions will be made to withstand robotics use. You do have a great eye for the parts that would need to be altered for real use. Will post a robotics ready version soon, I'd really appreciate it if you'd test that. Thanks for your continued support.
Open file (275.17 KB 1024x768 Chii_ponders.jpg)
>>17090 >I'd really appreciate it if you'd test that. Unfortunately, my printer isn't accessible to me atm. I'm sure several other anons here could help then? I myself should be able to as well at some point. >Thanks for your continued support. You're welcome Anon, but in your case it's more than just that. I think very highly of you, yourself. I believe you're an Anon that will very likely combine both a) succeeding at completing a workable humanoid robot that can serve as a nice baseline robowaifu prototype model, and b) doing it in such a way that maintains the dignified DIY, open-sauce, giving spirit that exemplifies the Christian nature of generosity. I hope we anons here will set the pace for the rest of the world in that regard concerning robowaifu development. You've come far already--we all have. But, we're still just in the foothills of this great mountain ahead of us, so stay strong & stay focused. Together, we're all going to make it! :^)
>>16849 Brilliant design approach, structurally-speaking Anon--very birb-like tbh. Birbs are the natural icon of lightweight+strong biostructures we can all learn from ofc. Old Leo Davinci (& literally thousands of other great men) studied birb bone structures thoroughly, and I think we all should too!
Open file (72.66 KB 461x634 grid-1-001.png)
>>16962 So, I'm working a bit on a internal structure that can be printed, be bend, and squished, while also holding tubes and cables. Not sure if it will work out, and how long it will take. We'll see. It's trial and error while listening to some (mostly doomer) stuff on YouTube or podcasts. I try to make it lightweight but also very controllable for how it feels and looks when it bends. That's why I'm not just relying on the patterns in the slicer. I want for example to make structures which are more uneven than a regular infill created by a slicer. This is needed for: - Prototyping - The cheaper variants - Maybe for all variants as a basic structure, to hold sensors, tubes and cables.
>>17110 Very interesting idea. This would also allow us to print soft robot breasts in the future as well. Hard plastic breasts would be fun for a while but I think a transition to realistic breasts would be needed sooner or later and your research could be very useful for it.
>>17110 you're on the verge of metamaterials right here I love it, keep up the work!
>>17110 Amazing looking. Godspeed Anon.
Open file (27.32 KB 461x634 grid-3.png)
>>17113 >print soft robot breasts in the future as well Maybe, but I consider breasts to be less of a problem. For example, they don't need to support anything. I wouldn't rule it out, but at least small breast could probably just be molded out of silicone. >>17114 >metamaterials I was looking into origami and something similar as well. This might be an alternative or addon: https://www.youtube.com/watch?v=B-x-1DrabYo and https://www.youtube.com/watch?v=_tKI8BUHFLo
Open file (52.27 KB 433x634 grid-3-004.png)
Open file (38.57 KB 433x634 grid-3-003.png)
Open file (196.12 KB 1920x1050 grid-3-002.png)
>>17117 Still on it, but I went with the idea of cutting wholes from the side as in the picture before. I'll try to print it soon. That's gonna be fun, lol. The final version would rather have a shell outside, so it would be easier, but I might try it without. It's supposed to be similar to the version on top of the picture in >>16849, but with additional lateral holes. Though, those would one be on the inside and weakening the structure there, the outside should still be closed (or maybe as an alternative, with very fine holes) and also have a structure that might be palpable through a layer of silicone (and maybe additional power mesh) and give the impression of muscle fibers. The final version might be less ambitious, it's just me exploring the space of possibilities.
Open file (26.61 KB 433x587 nose-T01v2a-001.png)
>>17121 I plan trying to print the "metamaterial" tomorrow. It's going to take more than 8h, if it works (which it wont at the first try). I might also print the Paradox Drive from >>17086, if I'll find the time. Worked a bit on the nose today.
>>17144 Looks almost alive. Or like it was made by spiders.
>>17151 i guess if you shape these right you just need rubber bands to piece a body together without having to care about moving or rigid parts actually i think ligaments are really just rubber bands anyway
>>17152 sponge-waifu when
>>17152 just wanted to add to this because its been on my mind all day supposing the joints are elastic then you could literally just make a marionette except have the wires be internal all converging in the torso like a piano, you would only need 2 pairs of perpendicular wires to get full 3d motion for each appendage, i tried to see how difficult it would be to program since your using tension as movement instead of torque but its really not that much different as long as you have something in between to tell you the relative angle, the rest like dimensions, weight and moment of inertia dont change so moving an appendage to a coordinate like (1,1,1) would just require calculating the required horizontal and vertical tension based on the angle, but im ignoring eccentricity for now to keep it simple not sure how this would work when picking up things since this would change the moment of inertia and the coordinates would misalign, although ive picked up things before that i thought were heavy and hit myself on the head, i think lifting is a different kind of process than just normal movement
>>17168 I don't think I can follow your thinking completely. However: - I generally think precise calculation of movements is a bit overrated. We only need to guesstimate it, start moving, and then measure how it's going. Don't forget about the sensors. Then the resulting data could be learned by a neural network to make better assumptions about how the body would need to move, if something changes, like carrying a weight. - The real issue with any primarily spring or "rubber band" based actuation is that the body would be under stress if resting in the wrong position. It would always want to snap back, and require something to hold it in place while the material trying to push it would be under stress. No only for that reason there might be problems with deterioration of the material. - This whole discussion should be moved to >>406 or meta / general engineering if we go on with it.
>>17169 i said rubber bands just as an easy way to have a flexible joint that can move freely, maybe a ball socket would be better, the wires would obviously always be under tension you will need a baseline tension of whatever the bending force due to gravity is so that it doesnt flop around when anything but vertical, movement being done by tightening one side while proportionally slackening the opposite side which gives a net force to the tighter side, both sides still under tension though its just a matter of shifting the balance between left/right and up/down wires and translating it into a 3d coordinate using the dimensions of the appendage to get a predictable model for movement you wouldnt need to actually calculate this in realtime, just once for every angle and tabulate the corresponding tension and displacement, its only when weight changes you need to recalculate since this changes everything, its a simple calculation that cant take more than a second to do all angles but i dont know how you could possibly recalculate if you dont already know the new weight
just one more post until its moved i think i figured out the lifting problem, its actually really simple, you would just increase the baseline tension and lift a little until you see the object lift ( which is the force needed to actually lift it), subtracting this from the starting tension would literally just be the weight of the object, with this you can calculate a new table to move with the object
>>17171 Yeah, it's very interesting. Sounds like one would imagine the tensegrity in the human body. That's probably how we should think about any kind of muscle type actuator, or of the motor is not in the joint but somewhere else while the joint is controlled through wires.
Open file (24.17 KB 532x634 grid-5-004.png)
Open file (22.06 KB 532x634 grid-5-003.png)
Open file (26.77 KB 532x634 grid-5-002.png)
Open file (19.09 KB 532x634 grid-5-001.png)
Open file (39.70 KB 532x634 grid-4-002.png)
>>17151 Still looking for good ways to design squishy or flexible parts. It will most likely be open cylinders which thinner and thicker parts. I have to order them in a way that this can be printed without support material.
Open file (891.12 KB 1080x810 IMG_20220818_011307_0.png)
>>17195 So, unfortunately, finding a way to design a model for some kind of flexible tissue that can ne FDM printed turned out harder than I thought. However, I think I'm on the right track now. I'll certainly need to do more test prints. I probably won't go with the idea of 'shooting' holes into a design from all sides, but build up a design. It's certainly still ways to go and the approach might fail. We'll see.
>bump for nice content
>>17242 You should either print varioshore tpu at full foaming as per the cnckitchen video: https://www.cnckitchen.com/blog/testing-colorfabb-varioshore-tpu-foaming-3d-printing-filament?format=amp Or, do what makes the most sense and cast silicone into a printed mold. The tpu is pretty flexible and soft compared to PLA, but much harder and rigid than silicone.
>>17565 This model is really very soft. Printed with some cheap chinesium TPU from AliExpress.
>>17577 >TPU >”guyz its really soft” Anon....
>>17578 Depends on the TPU but Ninjaflex 85A looks very flexible. But the guy is probably right that its better to print a mold and cast silicone in it rather than printing soft out of TPU. https://www.youtube.com/shorts/2REOKW-gWQw
>>17578 It is soft, maybe you don't realize that the biggest parts there (>>17242) are less than one millimeter thick. And there are empty spaces in between, which was the whole point. >>17580 It's not about silicone rubber vs TPU in general, but about combining those two. Keep in mind these a subcutan (under the skin) parts, not for replacement of a silicone rubber skin, at least not in the more advanced builds. Maybe certain parts could also be casted in harder silicone rubber later, or printed on a printer using that. However, I don't have such a silicone printer right now, and I want to develop and showcase RW's with different levels of difficulty and price. More expensive tools or more work will result in a higher price and easier gatekeeping. TPU still softer than a hard shell, btw. Also, it's simply useful for prototyping and and for creating internal bubbles or holders for sensors and other use cases. If I had to do that prototyping with silicone rubber, then you could add 10-20 years of development time. tl;dr: I will go on using TPU for prototyping, internal parts and wherever I see fit.
>>17586 >combining those two Tpu is a nice idea to make a structure inside of the skin, I think I understand what you mean now. It does not chemically bond to silicone, but does not inhibit the curing. I use tpu filament to form channels in my silicone castings that can be filled with conductive grease for a capacitance/resistance sensor. It pulls right out, and I think you could run into a problem if you try to bond a thermoplastic with a thermoset silicone. Making a loose skin "shell" of silicone around plastics would be like putting rubber gloves around a robot hand. The way I'm planning to use soft materials is to keep them isolated to few body parts and the rest of the robot would be an open hard shell. >printing tpu to make a deformable shape I have had limited success using one layer perimeter with a gyroid infill at a small percentage. Just model the shape and the gyroid infill takes out the guesswork for a mostly internal structure. The outside layer will carry most of the strength as you know though. As a matter of fact, why not just print the gryoid infill of tpu without any top/bottom/perimeters? I bet that would satisfy all your needs for this project.
>>17592 Not him. >gyroid infill nevery heard of it before your post. Would you please explain it further Anon?
Open file (328.85 KB 1339x1179 media_Ec-T3gwU0AAhjMO.jpg)
>>17592 I know silicone doesn't bond to anything, not chemically at least. But it can form a physical bond with a complex structure and fuzzy surfaces. In regards to the loose skin shell around harder parts. Well, that's vaguely how the softer tissue on top of the muscles feels. Of course it needs to be somewhat connected but it's totally good to have it a bit floating. I'm also more and more thinking of nylon meshes and fishing lines integrated into these internal parts. But that might be something for a later optimization. I want some semi-realistic muscle movement, at least in the thighs. >gyroid infill I've read about the method with using gyroid infill before but didn't think of using it here that way. I'll look into it. Same for only printing the infill without the perimeters or top/bottom shell. Might be helpful to some extent at least. Thanks for the reminder. The poblem will most likely be that I want it to deform three dimensional, not just vertical or horizontal. >>17598 Well, you literally just pick gyroid infill when selecting the kind of infill you want for your print in your slicer.
>>17602 >Well, you literally just pick gyroid infill when selecting the kind of infill you want for your print in your slicer. Fair enough. I admit I was rather hoping for a discussion of it's strengths & weaknesses--say from the mechanical-engineering perspective.
>>17603 Then it was a misunderstanding. Just picking some specific infill would be easier than changing the part in the modelling program and trying to make it more flexible. However, the infill would most likely go from bottom to the top like a wavy wall, with the waves going horizontal. Which could be fine enough, but it probably wouldn't be easy to squish the part between top an bottom. Also, it wouldn't be about designing internal spaces for keeping tubes or sensors at a certain place. It might still be useful though, especially for prototyping or simpler and cheaper variants.
>>17604 Same Anon. I actually looked at it now and it's better than I thought. I didn't picture it that way and didn't even try.
Open file (871.17 KB 877x956 tpu_tissue_test-01.png)
Open file (1.01 MB 908x923 tpu_tissue_test-02.png)
Open file (878.22 KB 1080x819 pelvis_openscad.png)
>>17602 Started working on a new pelvis. Also, test printed some TPU parts. Photos only, since my other device has no internet right now. I decided to work on a simplified hard shell robowaifu, using my files. She will only stand on the desk, have no lower legs and look like she's wearing a dress around the thighs. I'll focus on some rudimentary movements (arms, head, maybe hip). I want to see something working at the end of the year, but my main approach is slower and I don't see anyone else building something. So I decided to actually build a simpler variant. Printing, especially PLA, is actually so cheap that I shouldn't really care. I probably also have enough storage for failures and gradually improving demonstrators.
>>17630 >She will only stand on the desk, have no lower legs and look like she's wearing a dress around the thighs I'm currently considering something similar for Sumomo-chan once the basics of her system automation software are worked out. We can always hack her robo-bust onto a roomba's components later for some basic mobility. But for now just having a talking & expressive face & shoulders, then later arms & hands, will be a nice step forward and will force solutions to several fundamental control problems.
>>17630 This structure reminds me of adipocytes (fat cells) in a human body. Fat is certainly squishy...perhaps you are onto something here, anon!
>>17632 Neat! It really does look like some kind of cellular membrane idealization, SophieDev. Good eye.
>>17630 I like how the structural integrity of your design isn't particularly harmed by local errors/dropouts. Very resilient approach Anon!
Open file (2.72 MB 1560x2104 IMG20221109170426.jpg)
I've been working on my first design, a moebot (~60cm height), and in order to fit the shoulders in particular, I've had to work on a custom servo design to get the torque I need, rather than use off-the-shelf servos. Using a planetary gearbox + wormgear, I'm planning to gear down a coreless motor spinning at around 40,000 rpm to around 60 rpm. This is the only way I could see to get the torque I need in the space I have. However, I have some issues with friction in the printed gearbox. It's loose enough to turn with my fingers, but stiffer than I'd hoped. Also, I'd really appreciate any advice on the best methods for attaching worm gears to the drive shaft of the motor. I've tried super gluing the shaft into a hole in the worm gear, but it flies off immediately as the motor spins up. I have some stronger epoxies, but I don't imagine they'll be much better. Also, if anyone here has any experience with high speed gearboxes, is this design feasible? At 40,000 rpm, will the gears function in plastic? Should I try and find a way to have them machined in brass or aluminium? Thank you in advance for any advice.
>>17638 I wish I could be of more help Anon. Sounds like you're already doing things I might think of doing. Godspeed with your effort Anon, it's a very interesting design. About what size is this whole assembly btw for reference?
>>17638 >I've been working on my first design, a moebot (~60cm height), ... Do you have any prior experience? Two-wheeled balancing robot, quadruped, etc.? I hope you do. A humanoid robot of any size is no joke. >coreless motor spinning at around 40,000 rpm Hate to say it, but that motor is trash. A toy. The weight to power ratio is abysmal, and there is no way you are going to be able to grip that smooth little 2-3mm shaft hard to enough to extract any kind of real power out of it with a gearbox, especially with plastic gears. Typical applications of these little brushed motors will couple them to a gearbox by press-fitting them (with a vice or hydraulic press) into a brass gear. All around just not a good solution. >This is the only way I could see to get the torque I need in the space I have That's because you haven't looked into brushless DC motors, or you have and decided to avoid them due to cost and/or complexity. There is no avoiding them in mobile robots; they are state-of-the-art and the de-facto standard. >However, I have some issues with friction in the printed gearbox. It's loose enough to turn with my fingers, but stiffer than I'd hoped. That's because worm gears are garbage, I can't even imagine 3D printing one. High friction and non-backdrivable. Back-drivability is critical to your (our) application. Without it, expect violent disassembly on the slightest impact or resistance to movement, where back-drivable actuators will convert the opposing force back into electricity through the regenerative breaking effect. Look into all that if you haven't already. Unless you're talking about the planetary part of it, then I'd say it's just a matter of tolerances and designing around them. Most of the sexy printed gearboxes you see out there on the 'tubes will "cheat" by either printing in high-resolution and smooth, but weak and brittle, resin or print in proper high-strength nylon with a shit-ton of post-processing and hand-fitting. Also, I can't see the other side of your gearbox but, those planet and sun gears are going to need bearings for them. See the opentorque actuator linked below. Plastic on plastic is a big no-go. >Also, I'd really appreciate any advice on the best methods for attaching worm gears to the drive shaft of the motor. I've tried super gluing the shaft into a hole in the worm gear, but it flies off immediately as the motor spins up. You've discovered what I wrote above then. Solution: don't. >Also, if anyone here has any experience with high speed gearboxes, is this design feasible? At 40,000 rpm, will the gears function in plastic? No. Heat from friction will absolutely destroy it as well, if that motor has the output power to even do so. A lot of 3D printed plastics will start going soft on you at unacceptably low temps like 50-60C. But don't take my work for it, I'm not a 3D printing expert. Lurk around the odrive community or MIT cheetah or somewhere else that people are printing high-performance actuators and see what they're doing. >Should I try and find a way to have them machined in brass or aluminium? Since the design as it stands is non-functional, save your money. Some links that may help you: ht tps://yewtu.be/watch?v=VsecsMZuupI&listen=false ht tps://yewtu.be/watch?v=EhJa8kdkzRY ht tps://github.com/G-Levine/OpenTorque-Actuator ht tps://discourse.odriverobotics.com/t/opentorque-project-new-design/8280 ht tps://www.omc-stepperonline.com/planetary-gearbox?sort=p.price&order=ASC&mfp=300-suit-for-motor[Nema%2017]
>>17638 Another important thing that I forget to mention: BLDC motor KV ratings. KV is a motor constant, how many thousands of RPMs per volt applied, at no load. So a 2000KV motor, run at 1/10th duty cycle PWM off a 20V power supply, will spin at 4000RPM, unloaded. Sometimes these RPM figures get so ludicrously high that I wonder if the motor bearings will even hold up to being spun at max power unloaded for any amount of time. Probably not. I wouldn't try. But RPMs aren't what we're going for here, it's torque. For torque, lower KV is better, however unintuitive that may sound. You will notice too when shopping around that the lower the KV, the larger the size. KV<500 being >40mm diameter, KV<200 being at least 50mm diameter, and sub-100 KV motors being sized for scooters and those electric parasail things. This is unfortunate for the small to medium sized robot maker, but unavoidable. I can't give you a definite figure because I'm still in the experimental stages with motors myself but, my research shows that you want sub-100 KV for joint actuators. Keep in mind this is all direct-drive. There is also quasi-direct-drive for plebs like us: low KV motors geared down through back-drivable gearboxes like planetary or cycloidal (or harmonic, lol $$) and/or timing belts and pulleys. If you are just prototyping, and some or all of size/weight/power/etc. can be sacrificed to get a proof-of-concept up and running while you still have the motivation, look into stepper motors and micro-stepping controllers for them (e.g. Trinamic's TMC2xxx) combined with either planetary gearboxes if you need a reduction ratio of say, 1:5 to 1:30 (the higher you go, the less back-drivable) or, timing belts like a GT2 with 1:3 or 1:40 sprocket ratios. Dead simple to program for and come with built-in positioning as long as you aren't over-driving them. See links below. ht tps://www.trinamic.com/products/integrated-circuits/details/tmc2130/ ht tps://www.omc-stepperonline.com/economy-planetary-gearbox?sort=p.price&order=ASC ht tps://www.ebay.com/itm/334592608461
>>17644 >1:3 or 1:40 Meant 1:4. Pulleys can be found easily and cheaply with 15-20 teeth on the small end, and up to 60-80 on the large end of the spectrum, giving you a 1:3-1:5 ratio or so, depending on how large of a diameter you can tolerate on the larger sprocket (a 60T GT2 sprocket is ~42mm diameter),
>>17644 Thanks for the great inputs, Anon! :^)
Open file (92.44 KB 1000x1000 gimbal1.jpg)
Open file (86.42 KB 900x600 gimbal2.jpg)
Open file (74.31 KB 680x2500 gimbal3.jpeg)
Open file (120.32 KB 606x369 gimbal4.png)
>>17638 You could also look into gimbal motors if you can't afford the weight (> 500g or so for the above motors and planetary gearboxes). Gimbal motors will give you smooth turning and high torque in a compact size and low weight, at a cost of reduced power overall (Watts). Think high-efficiency hobby servos, if you pair them with magnetic absolute position sensors (AS5048, AS5047, AS5600, etc.) and magnets. Search for Ipower, T-motor, Emax, and Tiger brand gimbal motors. But good luck, because apparently all of the higher-power, low-KV models are out of stock, everywhere. They will still need to be driven in the same manner as regular 3-phase BLDC motors but, due to their lower power consumption, you can roll your own controllers on the cheap (like 15USD BOM) using motor drivers with built-in MOSFETs like TI's DRV8311 or their other DRV83xx chips with cheap external MOSFETs. ht tps://www.ti.com/lit/ds/symlink/drv8311.pdf ht tps://www.ti.com/lit/ds/symlink/drv8328.pdf
Open file (2.45 MB 4624x3472 prototype.jpg)
>>17643 >backdriveable motors >print high strength nylon >bldc with planetary gearbox Holy shit are you me? Finally, I can show my prototype to someone who can give me advice: >pic related It's a moteus controller with a 8318 bldc. I have not tested it but it's completely assembled and needs to be calibrated by pre-existing software. I want to make nice, quiet, and durable gears but right now my best option is nylon planets with either polycarbonate or metal ring and pinion. I am wondering about using gear cutters to post-process printed gears since they have strings and warping. I am a little bit cautious to jump into herringbone gears since I have not made a gearbox prototype that I like yet. >gimbal motors Do these have enough power to move around 20-60 lb of bot? >control scheme So you have done motor research but how will you control the robot? I am thinking of adapting the moteus system to 3-4 joints driven by various sizes of motor/gearboxes and pulleys/belts. Is there a support community as large as ordinary high amperage BLDC for gimbal servos?
>>17679 >Holy shit are you me? Everyone usually arrives at the same place with just a little research. BLDC motors, for all their drawbacks and idiosyncrasies, are the only option for mobile robots are this point. >It's a moteus controller with a 8318 bldc. That's some serious kit, anon. I believe the MIT mini cheetah uses 8xxx motors too, just heavily modified. The large size lets them fit a custom planetary gearbox in the space inside of the stator. Check those actuators out if you haven't already; I believe they're open source. >gimbal motors >Do these have enough power to move around 20-60 lb of bot? Absolutely not. That is the biggest downside to gimbal motors: they sacrifice total power for low end torque. Compared to a motor like yours, with phase resistance in the milliohms and total power of probably 2000W+, gimbal motors will have a phase resistance of 10-15 OHMs. This, combined with the lack of back EMF due to low or even 0RPM speed in their usual application, makes them more-or-less follow Ohm's law, where power will be limited by how much heat you're able (or willing) to dissipate in the coils. Think 5-20 Watts with a 3-4S LIPO pack. I made the suggestion to the anon above because he was already trying to use a small motor with no real power or torque. For example: ht tps://shop.iflight-rc.com/index.php?route=product/product&product_id=279 > Specification: >Model No.:GM5208-12 >Weight:229.2 >Dimension(mm):63*22.7mm >RPM:456~504 >Torque (g):1800-2500 >Voltage (V):20V >Cunnrent (A):0.09 >Configuration:12N14P >Resistance(Ω):15.2Ω±5% I don't know where they get that 0.09A figure from but, 20V / 15.2R is 1.316A, or 26.3W power, stalled? Not a lot of information out there, will need to test and characterize yourself if you're trying to use these. I'm going to be picking up a couple for this purpose, to see what they're capable of. Not gonna claim to be the gimbal expert at this point. Also keep in mind that this is on the larger end of gimbal motor size. 22xx, 28xx, etc. are likely beneath consideration for anything besides spinning a LIDAR around or similar. >control scheme >So you have done motor research but how will you control the robot? Right. I'm still at the planning and benchtop testing stage like yourself. I haven't given much thought to higher level control yet, only that I'd like velocity, position, and torque control to start with. From there? The usual inverse kinematics using the motor position sensors for joint position sensing I guess. >I am thinking of adapting the moteus system to 3-4 joints driven by various sizes of motor/gearboxes and pulleys/belts. Not sure what you're referring to with "moteus system". Do you mean the combination of a BLDC motor and the moteus controller? That's what most people are doing, yeah. Often one controller to one motor, then connecting them all to a central controller using CAN bus over CAT-5 cables/RJ45 connectors. I don't know if I'll be using CAN or RS485. Both are noise-resistant and high enough bandwidth for most control and sensing applications. Modern micros though, will often have 3-8 UARTs and only one CAN peripheral. RS485 should be the cheaper and simpler implementation, but eh. Who knows at this point. >Is there a support community as large as ordinary high amperage BLDC for gimbal servos? No, not that I've seen. Apparently brushless DC gimbals themselves are a fairly recent innovation. I don't have a link right now but I have seen discussion going on out there. For those curious though and want to dip their toes in on the cheap, look into SimpleFOC controllers. The SimpleFOC firmware can be flashed onto a number of cheap 2-3 axis gimbal controller boards, giving you control of 2-3 motors for 10-30 bucks and not the 200-500 you'll drop on the same number of moteus and odroid controllers. RCTimer too, apparently has stock of gimbal motors ATM (not an endorsement, I've never ordered from them.) Worth checking out. ht tps://docs.simplefoc.com/bldc_drivers ht tps://rctimer.com/gimbal-motor-c0428 ht tps://rctimer.com/rctimer-2-axis-brushless-gimbal-controller-amp-imu-v10-p0444.html ht tps://www.ebay.com/itm/174343022855 ht tps://www.ebay.com/itm/174342062986
Open file (147.93 KB 1620x1080 minitaur.jpeg)
>>17680 Forgot my pic. Ghost Robotics' Minitaur. Note the text on the motors (unmodified and direct drive?) and the integrated controllers.
>>17682 Right. The oldest two labeled "Ghost Robotics Early Direct-Drive Prototypes", from before they went full "killer dog bots: now in Coyote Brown" trying to get a turn at the MIC infinite money spigot.
>>17683 Yep, I kinda figured as much. TBH, it will be no different for our robowaifu designs once they are effective. The need for robowaifus is plainly a clear one; almost ironically somehow, the very same Globohomo groups that created that need to begin with, will also abscond with our and others' robowaifu work to make bombs, obvs. *sigh* I don't think we can 'have our cake & eat it too'; either we have open-sauce, freely-available robowaifus for everyman + clownworld killer waifus & dogs, or nothing only clownworld killer waifus & dogs. *sigh* >=== -minor reality adjustment
Edited last time by Chobitsu on 11/15/2022 (Tue) 22:46:51.
>>17683 here's a compilation of similar quads during current year. https://www.youtube.com/watch?v=dwgg4vd9TII
>"There’s also no gearbox, and that offers a number of advantages in terms of the robot being able to feel forces on the legs very quickly, and then also react very quickly."[1] So, how do they do that w/o sensors of some sort (AFAICT, they said they don't have positional sensors that can break)? Measuring the voltage drop/changes across all the actuators and then inferring the changes that way, or what? 1. https://spectrum.ieee.org/ghost-robotics-minitaur-quadruped
>>17690 >So, how do they do that w/o sensors of some sort (AFAICT, they said they don't have positional sensors that can break)? I believe they are referring to the lack of any kind of gearbox or belt-induced backlash here. I guarantee you they are using position sensors though; you need them for BLDC commutation, whether that's absolute position or electrical (hall effect). They're inside those black boxes on the motors, probably a contactless magnetic rotary position sensor from AMS mounted to the controller board like on the moteus. This line, however, >Again, all the sensing is through the motors, there’s no current sensors or force sensors. I find hard to believe. Low side current sensing with H-bridge configuration drivers is trivial; a low-Ohm current sense resistor, an op-amp, and an ADC will do it. Most BLDC gate drive ICs (DRV83xx, etc.) have this built in. You're gonna want it for current-limiting for safety at the very least. Force sensors? Sure. You can sense impacts and obstructions by comparing expected position, velocity, and power (needs current sensing) versus actual. >Measuring the voltage drop/changes across all the actuators and then inferring the changes that way, or what? Outside of sensorless rotational speed sensing in brushless ESCs, this isn't very accurate or useful, because of back EMF. They're definitely position sensing and rigid body modeling. Back-flipping and door opening too, that shit's hand-tuned like a Blender animation.
>>17694 I've read that the mjbots quad a1 (prob uses a moteus controller) measures phase currents in a current control loop to detect electromagnetic torque. It doesn't detect cogging torque, gearing stiction, or friction. To measure all mechanical torque, you would need a seperate torque transducer. I'm not sure what kind of torque transducer I'd want to have.
>>17702 I don't recall ever seeing active torque sensors in recent, popular robots. I could be wrong though, Atlas probably has a bunch up in that million-dollar chassis somewhere. I was under the impression that the torque is measured in the early design stages on the bench, usually by holding the motor in a mount and measuring the relevant operating conditions (voltage, current, angular velocity, coil temperature, etc.) as progressively heavier loads are moved by an arm on the motor shaft. The torque then, would be derived from those measurements in real time in the final application. Friction and such would already be accounted for using this method. Cogging torque is a little different though, as it's dependent on rotor position as well as direction of rotation. This, too, can be measured in the motor characterizing though, and compensated for using a look-up table. Look into software anti-cogging or cogging torque compensation. An article about it: ht tps://hackaday.com/2016/02/23/anti-cogging-algorithm-brings-out-the-best-in-your-hobby-brushless-motors/ Note that this is from 2016 and pre-dates a lot of the popular robots using commodity BLDC motors out there, and probably BLDC gimbals as well. Studies like these ht tps://www.modlabupenn.org/anticogging/ are likely some of the foundational work that made those possible. As far as your torque sensors go, I'd recommend trying to characterize the motor as best you can without one first, using the easy to measure voltage, current, rotational speed, etc. (see above) and see if that's sufficient for your application. For instance, seeing if the motor you want to use even has enough torque to move the expected mass of your robot (perhaps divided by X number of motors if it's bipedal or quadrupedal, etc.). Or, if you're trying to do something fancy like jumping, if the motor has enough angular acceleration/velocity at that torque. Keep in mind that angle of the force vector? or whatever needs to be taken into consideration in your torque calcs; it's a lot easier to move a weight fixed to an arm from the 5 o'clock position to the 7 o'clock than it is from the 8 o'clock to the 10 o'clock. Need to integrate over the angle moved iirc. Or you can use an arm pushing on a load cell, where the distance moved is 0 and ignored. Also gotta keep in mind margins, losses in drive trains, external forces, etc. But whatever: just get that motor on the bench and get yourself a data-logging setup. Bolt the motor down to some wood, clamp it to a table, make a test arm with a flanged shaft coupler and some aluminum extrusion and go nuts. What I want to say to everyone who asks if this or that motor is good for x, y, or z. Fuck if I know, you gotta test it. I rambled a bit but that's my 2 cents on torque.
>>17705 great inputs Anon, thanks.
>>17705 >torque I've thought about this and I've decided to just test the motor to obtain a max torque value so the motion control scheme has a figure not to exceed. No need to overthink it in the design stage. >cogging Yeah I've noticed this too as soon as I had a motor in my hand. I went with the moteus because I had a feeling the dude would make a cogging torque compensation at some point, and luckily the boards after revision 4.11 have anti-cogging and some more secret sauce that my mech engineer brain doesn't understand >writeup https://docs.google.com/document/d/1M4O-LVki03qonNTkydv1N8gRteGIZk3ZyHd-mBpxL-U/edit#heading=h.x1qddq7g5cts >torque sensing setup I will probably use a weight scale and block of wood. Overcomplicating the project in the design phase is a rookie mistake that kills projects. The fastest way to the finish is to iterate on an achievable first version and only solve the problems that present themselves along the way. But you already knew that, didn't you? >moteus shilling I'd take a cheaper controller and motor if I could get large smooth torque and power output. I have thought about making a highly powered hydraulic system with a central pump or artificial muscles, but what I've came up with is not as safe and readily available as what I have right now. Maybe what we need is a less precise hydraulic system with a simpler motor to avoid the relatively large cost (1.5-3k) of controllers/motors. Something akin to big dog by boston dynamics? Their first robot to walk used hydraulics powered by a lawnmower engine. Pneumatics is out of the question because of it's need for a loud compressor and heavy tank. Nitinol is better off as kinetic energy absorbing material and shouldn't be shoehorned into a muscle, don't waste your time and money on it. Have you found any promising shape memory polymers? I haven't. This commentary about actuators may be better off in >>12810
>>17709 >This commentary about actuators may be better off in >>12810 True. I'll consider the move after it seems to me all the pertinent conversation dries up. If you think I've overlooked it at some point remind me in the /meta pls?
Open file (797.27 KB 1080x814 IMG_20221206_100356~2_0.png)
>>17630 >I decided to work on a simplified hard shell robowaifu, using my files. She will only stand on the desk, have no lower legs and look like she's wearing a dress around the thighs. I'll focus on some rudimentary movements (arms, head, maybe hip). I want to see something working at the end of the year, ... I started printing, but don't hold your breath. Every step in this endeavor seems to have some hurdle for me. I'm far behind where I wanted to be and it's hard to stay motivated to do more than just a little bit every day. It's getting better, though, now that I printed something. Picrel is just one half of the bottom and internal beam of the desktop waifu version. She's going to have hips around 25-30 cm. The hole in the middle will hold the plate with the motor tree. From there the servos should control some movements above through strings: Back-forth, sideway tilting and rotation. From the sides an enclosure formed like a skirt will be able to be slided on. The whole idea here is to have some rudimentary robowaifu, and keep areas separated for now. I don't need a full nody to work on the head and upper body. Or on having a talking head with voice recognition. The hip-leg mechanis can be worked on separately. Making a version with soft material, a spine and bones will also take way longer. Hard shells are easier to do. Though, even that assembly will be made out of parts that I could replace later with softer ones.
>>17998 >and it's hard to stay motivated to do more than just a little bit every day. I find that way is actually pretty relaxing Anon. Just keep moving forward is the key. :^) >The whole idea here is to have some rudimentary robowaifu, Super-excited to see what you manage Anon, Godspeed. I epect robowaifu 'busts' is probably where most of us will start for our first IRL bits tbh.
Anyone have any biology or chemistry rated projects that might be useful to our goals? I have a biochemical engineering degree so I don't have the skillset for the Mech E or AI related projects.
>>18009 What about giving robowaifus a sense of smell or taste? Think BioChem maybe has something to bring to the table for robowaifus in that way Ribose? Also, I posed the question about possible conceptual similarities to the Deep Docking approach to data filtering going on, and our own needs to pare down data as much as possible for our robowaifus low-end runtime hardware (>>18006). Obviously, this domain of molecular biology is extreme Blue Sky stuff before all is said and done regarding our robowaifus, but that's at least decades out. 'Home Petri Dish Onboard Biolab' is probably closer to our speed r/n haha. :^)
>>18010 Biosensors would definitely be the way to go for taste and smell. I know oyster mushrooms and slime molds have been successfully used as biosensors. There would certainly be a lot of engineering challenges, but crude mushroom biosensors might actually be doable for a proof of concept. A full biological system for taste would need a means to dispose of waste and to keep the organism alive without it growing out of control or fouling. I guess I'll get started on growing oyster mushrooms.
>>18012 Thanks for your interest. We have the Cyborg and Biology thread for that topic >>2184. The 'prototype and failures' thread should be used for showing off prototypes to report on progress or failures to learn from. Anything biological is probably only going to be used in the more ambitious projects or way down the road, also only if it's better or more accessible than something easier to handle. The most likely use of biology will imo be to use water kefir in the saliva. One way to keep germs at bay and help clean small cavities in the mouth. These Kefir christals are also something that could be shredded down into a liquid that could then later be disposed off while being in the bathroom. Keeping the orifices clean will imo be the most widespread use of chemistry and biology, aside from the fact that everything is somehow chemistry. Further discussions, exchange of options or speculation on this should imo be done in >>2184.
>>18012 Yes, any time we intentionally add biologics into our robowaifu's systems, we're probably adding at least 10%-15% add'l D&E R&D to the final workload I'd estimate off the cuff. We'll certainly need to keep the little buggers alive & healthy & watnot. > I guess I'll get started on growing oyster mushrooms. Sweet! Biocomputing here we come! :^)
>>18013 I can come back through several threads after the holidays and do some rearranging after the fact as well. If I forget, please remind me in the /meta thanks Anon.
>>17998 I'm just doing some testing today. Found some problems with printed parts which were sized down. I also confirmed that it's probably going to work, and also had some new ideas. This is the hip area, which will go on these internal poles from before. Inside is a mechanism which can tilt back and forth will also rotate 30° or so. The rotation will come from a pole, but the tilting will be done with strings, while the mechanism will additionally be connected to the body with rubber strings. It will be compliant, so I can move her around even without the servos, but she will move back to the position which is indicated by the servos. The upper body is supposed to tilt sideways on top of the pear shaped part, while also attached with rubber strings, so it can't fall off. In a later design, the tip of that pear might also be a wheel or even some kind of servo.
>>18025 Neat! Looks like it's a pretty compact design Anon. I'd suggest you don't be shy about 'hacking' around (as in the original back in my granpappy's day sense) with your early designs. Remember this is prototyping at this stage, don't fall in love with anything design-wise at this point. Very stoked to see your forward progress again Anon. Now just keep moving forward with it! :^)
>>18025 I had to play around with the model, found some flaws which I fixed in the 3D file, but the general idea seems to work. I drilled some holes into my prototype and made another version of the tilter, so I was able to test it using simple rubber bands. I will need bands to keep the mechanism in place as long as there's no force, while others tear it in one or the other direction. Better image quality this time.
>>18159 >using simple rubber bands Nice! Starting out with as simple and inexpensive components as possible at this preliminary stage is very wise. BTW, they make absolutely yuge rubber bands and really an astounding variety of them. Some are very tough indeed. Just look around for suppliers of a wide variety of higher quality ones I'd suggest. Good job Anon, keep going!
>>18168 Thanks. Yeah I saw these rubber bands or flexible ropes and kept them in mind, also therobotstudio on YouTube used ruber bands in the past. That's how I got to this point. For my most ambitious and long term project I thought using rubber bands wouldn't be a good plan, since they might wear out. Also, it could be a problem if the waifu would be at some point supposed to hold a certain position, but the rubber won't let her. For the more simple body it should work, though, and I also had some ideas to mitigate the problems in other builds. I generally think, having small servos which are adjusting ropes and change their force or it's direction, will generally be important for building a human-like robot.
>>17605 >https://youtu.be/tFgB0WqzkwE It looks like a company stole your concept, but they use resin-printed flexible struts.
>>18176 Haha, thanks, didn't watch it yet, but it's a kind of obvious thing. Didn't know we can print flexible with SLA, though.
I forgot to announce that my older files are on the Mega account I mentioned here: >>12661 I only update this from time to time though, also I will probably recreate most of my more bionic files in OpenSCAD or something similar. Link again: https://mega.nz/folder/nv42EbzT#1MUblavva2UG6DukgD0moA
Open file (468.46 KB 1046x2060 WorkInProgress.jpg)
Been exploring low cost and low mass methods for rapid waifu fabrication. This headless thing is slightly over a meter (a yard) in height.
>>18216 You are on the right track at this stage of developments. Just as I encouraged OSRM (>>11446) Anon, please look for every opportunity to use triangles in your design. They (alone) have the unique geometric characteristic that forms space-frames. Keep moving forward! :^)
>>18169 I'm still on it. Had some problems with my fillament being moist and this generally made me stall the development a bit. I'm really running into hurdles at every step. Though, I printed a bigger and more advanced model. I'm still close to have at least a simple model for the upper body with the required sideway tilting functionality. I'm also thinking about how to model a holder for the motors, which includes some standard metal parts from a shop, so I won't rely on plastic which would get soft at 60C. I wanted to use the screws on the bottom, but I can't connect them if there's a gap holding the metal plates, so I need to use the front holes or get a bit longer screws to reach the inserted nuts. The motor holder (tower) will be on the middle plate which is the gap in my small model in picrel. One servo to rotate, one for back/forth movement, one for sideway tilting. Stacked uppon each other. >>18176 So, after watching it: This is actually good news. Their software and approach might be useful to mass produce certain dolls and robowaifu variants. It's seem even to be available as a service to order parts.
>>18299 Nice! That's starting to look like it's getting to the ~145cm size-range, Anon?
>>18299 > I wanted to use the screws on the bottom, but I can't connect them if there's a gap holding the metal plates, so I need to use the front holes or get a bit longer screws to reach the inserted nuts. No idea what you're talking about here but, your second pic is a stepper motor (NEMA17?), right? You should be able to find all sorts of mounting brackets for them. If those don't meet your needs, you're better off cutting and drilling your own aluminum sheet versus trying to cobble together heavy and unworkable steel brackets from the hardware store. Aluminum is surprising easy to work with, even with just hand tools. Easy to fine-tune fit and such too, using files, etc. >The motor holder (tower) will be on the middle plate which is the gap in my small model in picrel. One servo to rotate, one for back/forth movement, one for sideway tilting. Stacked uppon each other. Not seeing it. Do you have a sketch or a CAD mock-up or anything to illustrate?
>>18303 >~145cm It doesn't need to, since this one won't be sex-enabled, while also having boobs like an adult. I guess even smaller versions would be possible, but I guess with smaller servos then. Anyways, the current plan is to make her with hips around 20cm. For a bigger variant, I'd need a bigger printer or make at least one more cut. Her body below the belt is going to be just a kind of dress-like enclosure which ends around the knees, which makes her shorter anyways. The design is mean to be on a chair or desktop. I can put a bra on her, as soon as she has nipples, if you're concerned. Gonna be fun buying such a small bra in a local store as an adult male (Chobits comes to mind...) >>18328 It's a NEMA 17, yes. I'm not going to make my own bracket or such out of aluminum, let alone for the reason everyone copying the design would need to do the same. I didn't see a bracket when buying it, and had the idea to print it in mind. But I just need to hold it somehow, not an enclosure. These stell brackets are for shelfs and cheeply available everywhere. I don't need to alter them at all. The brackets will hopefully hold the servo while they will rely on some printed parts. The metal parts are mostly about preventing the plastic from getting heated up. >Do you have a sketch or a CAD mock-up or anything to illustrate? I made one after you asked (picrel), and yeah I know it looks wild.
>>18329 >if you're concerned Lol no. You do as you see fit Anon, that's your affair. As you probably know we went over this before as a board community, and it will be good for everyone here once we can learn to properly manage the square-cube law, etc. >Gonna be fun buying such a small bra in a local store as an adult male (Chobits comes to mind...) Poor Hideki! That was a great episode. :^)
>>18329 My progress was once again much slower than anticipated. I discovered that my idea about the tilting of the upper body was very flawed, and I started to work on a solution for it. Also had some other stuff to do. I thought I just tilt the upper body in OpenSCAD in various directions, and substract the affected parts from the upper body. Kind of obvious that this can't work, since the body wouldn't have anything to sit on while being in neutral upright position. So I started to design and experiment with a optimized part for tilting the right way, but it turns out to be a bit difficult.
>>18400 If you could create some kind of animation of the tilting of the upper body you mention, and how the fail in neutral upright position happens it might help us to provide you some advice? Regardless, just keep moving forward Anon. That's what will count in the end! :^)
>>18401 Maybe I will create animations with OpenSCAD at some point, but it's not that easy and useful for that specific problem right now. Also, I just configured my PC to work properly with internet but still mainly posting from my tablet. I think I solved it. I was thinking in too complicated ways. Simply sliding sideways and a then tilting on the sides is good enough. It works with a extra sphere sticking out in the middle of the tilter, or using a propper pole going through the tilter, where there's a reverse U opening. The three variants on top of my drawing probably won't work or I don't like it, the two on the bottom will. The right one is easier and it's how the one in the picture works. Though, the whole complex mess of huge openings is completely unnecessary, but it helped me to test it. Back and forth tilting using the strings works like a charm, btw. Still, I don't think I'll make it working completely, with arms, head and shoulders, till new years eve. But I think I'll "finish" a working prototype with motors for moving the main joints in January. Probably still with me switching motors out for different functions, though. I only bought five Nema 17 and still have some drone motors for twisted string actuations, and some other small motors.
>>18430 Excellent, sounds like you got over a hurdle Anon. Good job! >The right one is easier and it's how the one in the picture works. For me your pictures are a little small for me to be able to zoom in and really get a detailed look at things. OTOH, that's a beautifully-elegant set of arches in the upper torso! How in the world did you manage to print that? :^) Looking good Anon, keep moving forward.
>>18432 The photo on my tablet isn't that great and there isn't much to see. It's just bottom right approach on the drawing. It just turned out to also not work very well, since of course it falls off when tilting back and forth. Only the tension from the plastic arches holding it, which isn't a good idea. The whole design has it's limits, due to be rather simple and using hard parts, which can't just move out of the way.
>>18433 Ahh, I see then. >The whole design has it's limits, due to be rather simple and using hard parts, which can't just move out of the way. Well, you obviously have enough smarts to figure this pivot/rocker thing out Anon. I'd suggest you just step back from it for a few hours or even a day or two. Go get a fresh perspective somewhere maybe. BTW, seeing you and the others all active here, for me, is like getting a Christmas present already! :^) Merry Christmas Anon.
Open file (956.80 KB 745x853 ClipboardImage.png)
>>18433 My brother in Christ, you're designing a motorized joystick waifu. This could be helpful for you. https://www.thingiverse.com/thing:4576634 I fully support your use of steppers, NEMA 17 for the win.
>>18440 Haha, that's not completely wrong. Especially since I want her to be compliant. Meaning, if someone pulls or pushes her upper body manually it will tilt and move without the servos and it will flip back after removing the force. I'll look into the joystick design.
Open file (15.73 KB 456x683 motor_tower-001.png)
>>18452 Still making progress. Picrel might be a bit better than >>18329, though it's unfinished. The cubes are the servos, obviously.
>>18477 Neat! I think Kiwi's advice is a good one. Keep moving forward.
>>18479 Yeah, I was watching the video. I was planning to use hall effect sensors anyways, I already have them. The mechanism itself is also interesting, but still a bit different from my approach and I won't just go for a complete redesign at this point. Maybe if my hip tilting mechanism doesn't work, I could switch to something like it. I still have servos in there, though, and they are far below the hip in my design. Also the joystick doesn't rotate around the hip and it doesn't slide sideways. I think, in a human-like creature some of the elements needs to move away from the sides of the hips if the body tilts sideways. Without soft parts, this is hard, but just tilting probably isn't the best way. Anyways, I'm still working on it, step by step, but there are always more challenges and details I didn't think of at first. So it will still take some time. I was also playing around with one of my face designs again, but it's still quite crude. It still works best when I sit down and listen to long podcasts with a not too attention demanding topic while working on it.
>>18489 >and I won't just go for a complete redesign at this point. That seems like sound advice IMO. Learn from others, but chart your own course Anon. If we all do this, then our creativity as a group will bloom! >but there are always more challenges and details I didn't think of at first In my experience, this is the way life goes pretty much always. Just take your time, but keep making little bit of new progress or new insights daily. Together, we're all going to make it!
>>18489 Currently working on the details of the motor tower. Need to separate the parts into pieces that can be printed and should clean up the code. More than 500 lines alone for that part of the SIMPLE body. Oouff. I'm not working too hard on it, though, and still listening to podcasts or videos while working on it. These little houses are for fairies, btw, or bugs maybe.
>>18548 I think you've packed the motors for 3-axis pretty tight now Anon. >500 lines It's the nature of what we signed up for here, you know. :^) Keep going!
>>18551 >I think you've packed the motors for 3-axis pretty tight now Yeah, we'll hopefully see if it works soon.
>>18552 My progress got stalled during the last weak, since I got other things to do, also needed a break. I started to do some testprints and had to fine-tune the model. I started working on other parts, to have a bit of change. I temporarily can't print and need to do some other things anyways. But I will go on as I can.
>>18626 >But I will go on as I can. That's the spirit, Anon. We're all enjoying your progress BTW.
>>18626 I'm conscripting you into the MaidCom project. We need another one that can do CAD work. I like what you're doing with your current project, though I believe it could be greatly simplified to meet your needs as a waifu joystick. Join me and I'll help your project.
>>18645 Thanks, last time somebody told me that I have to join military training I left the country until conscription got discarded, lol. Anyways, I have some things to do and otherwise I need to chill a bit. Also, I'm not done with my design. I just restarted doing some testprints. For the whole waist area I'm in the optimization phase right now. That means I'm close to the finish line, but the left over work might be a bit annoying. Of course our designs should be as compatible as possible to each other. I was thinking of importing your STL files into OpenSCAD soon, to check how hard it would be to swap out my chest design with yours. My work is to some extend more like a framework not a very specific design. If your idea was to look into the leg design, well I may do a bit work there, but I moved towards doing the upper part for good reasons: I might be to ambitious in regards to thighs and also for building something with an AI the upper body part is more important for now. I'm also stalling my new registration attempt in the forum again and again, but for joining your project I guess I should do this first.
>>18658 Welcome aboard Anon. Here's the final chest design for MaidCom: https://files.catbox.moe/0vltqa.stl. Glad we are both thinking of a framework. You'll notice her chest is rather spacious and designed for ease of adaptation. When you're ready, we'll get to work on a stationary MaidCom. You've opened my eyes to the value of a maiden that remains where her master places her. Though we can focus on her legs if that is your preference. Just know that we're years away from her walking. For now, rolling on skates is the best compromise.
>>18626 Still on it (picrel). We'll see if and where it fails, but I should be able to change it reasonably fast. Well, the code could be cleaner and be more adjustable, tbh. Still have to work on that. I also need to test the servos soon. Also integrating different chest variants, I guess. And finally printing the lower part in full size, and hope it's fine. >>18661 So what again would you like me to look into? Integrating your chest design with my current design? Yeah, it's a interesting chest, I have take a closer look on my PC.
>>18672 That seems to be shaping up into something pretty pro, Anon! Keep going forward.
>>18661 Started to play around with STL import. The merge looks interesting. But I hope you didn't think the tiddies would be where these openings are. It's too high, or I lack knowledge of the female anatomy. Anyways, thanks for motivating me to experiment around. Had to get off a game, I'm relaxed enough I hope.
>>18683 Unfortunately I couldn't really render this. It seems to not be a closed mesh. [Quote] In order to clean the STL file, you have the following options: - use http://wiki.netfabb.com/Semi-Automatic_Repair_Options to repair the holes but not the self-intersections. - use netfabb basic. This free software doesn't have the option to close holes nor can it fix the self-intersections. - use MeshLab, This free software can fix all the issues. Using MeshLab, you can do: - Render - Show non Manif Edges - Render - Show non Manif Vertices - if found, use Filters - Selection - Select non Manifold Edges or Select non Manifold Vertices - Apply - Close. Then click button 'Delete the current set of selected vertices...' or check http://www.youtube.com/watch?v=oDx0Tgy0UHo for an instruction video. The screen should show "0 non manifold edges", "0 non manifold vertices" Next, you can click the icon 'Fill Hole', select all the holes and click Fill and then Accept. You might have to redo this action a few times. - Use File - Export Mesh to save the STL. - If Meshlab can't fill the last hole then Blender might help: Start Blender `X, 1` to remove the default object File, Import, Stl `Tab` to edit the mesh `A` to de-select all vertices `Alt+Ctrl+Shift+M` to select all non-manifold vertices `MMB` to rotate, `Shift+MMB` to pan, `wheel` to zoom `C` for "circle" select, `Esc` to finish `Alt+M, 1` to merge or `Space` and search for "merge" as alternative Merging vertices is a useful way of filling holes where the vertices are so closely packed that the slight change in geometry is unimportant compared to the precision of a typical 3D printer Sauce: https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/STL_Import_and_Export#import_stl [/Quote]
Open file (166.97 KB 708x560 Manifold.png)
>>18683 The openings were not meant to match with her breasts. Your placement is generally correct. >>18684 It shows up as a closed manifold mesh in Blender, Cura, and my CAD programs. I don't use OpenSCAD though.
>>18689 Ah, okay. But is this the STL? Maybe it's only working as a Blender file. Idk. I should install Blender myself anyways. Also, it doesn't necessarily matter, I can still design parts that can be combined with it. I just can't render them together for printing, which means also no merging of parts in OpenSCAD.
>>18694 I imported the STL into Blender to make sure it was manifold, it was. All programs I use see the STL as being manifold and attempts to repair do nothing. You could try MeshMixer to combine the parts, I can combine it with almost anything else in there to create a new manifold mesh.
>>18700 Hmm, seems to be related to OpenSCAD then. I might look into it further at some point, but as I already wrote it's not that much of a deal anyways. I'm going to look motr into your MaidCom thread, reading about what you have planned. Maybe we should comment there if something relates to your design specifically. I hope you see with my merged design that you can't just add bulky motors on the outside of your chest. You would get masculine looking shoulders like Sophie had. I'm currently thinking about how to improve my own upper body design, the head tilting mechanism and how to do the shoulders and arms.
>>18700 Did you design this from the scratch in Blender, or rip the basic design off some game? It's the same old problem I had with SophieDev's files and some others where I cut something in a slicer and exported it. Basically; > - Render - Show non Manif Edges > - Render - Show non Manif Vertices > Remove... Then reconstructing it somehow... I did this so often and hated it since it lead to nothing, I won't get hugely into this again. This was the reason I gave up on importing STL into programs and trying to change it, until now. However, I just cut off a piece from a BJDoll design in OpenSCAD and still was able to export it (2nd picrel), though it shows errors in the preview. However exporting your design doesn't work. As I wrote before, it's not necessarily a problem, we just can't combine your design directly with something in OSCAD and export it as one file.
>>18739 Could someone else please test the file? The file was a .obj file made in MeshMixer, imported into DesignSpark Mechanical where I added the holes and made it an STL. It shows up as manifold for me but, I want it to work for others. Please other Anons, test this file. https://files.catbox.moe/0vltqa.stl . Also, I can export it as other filetypes such as .obj. What do you prefer OSCAD Anon? >=== -cleanup DL link
Edited last time by Chobitsu on 01/16/2023 (Mon) 03:05:27.
Open file (296.93 KB 1301x935 2023-01-15_19-11-41.png)
>>18747 >Please other Anons, test this file. Other than the issue that your scale is way off, looks fine in Blender 3.3.1 to me, Kiwi.
>>18748 How big is it? Should be 160 by 116 by 88 mm
>>18764 Could just be Blender's STL importer then. See those little squares down on the grid? Each one of them is 1 meter squared. So yeah, big as a house tbh. :^) >inb4 Kiwi's Robowaifu Giantess thingy >=== -add 'tbh' cmnt -add funpost spoiler
Edited last time by Chobitsu on 01/16/2023 (Mon) 07:01:46.
>>18747 >What do you prefer OSCAD Anon? I don't care, I have just started experimenting with imports into OpenSCAD. It supports: >> 3D formats >> STL (both ASCII and Binary) >> OFF >> AMF [Note: Requires version 2019.05] >> 3MF [Note: Requires version 2019.05] > https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/STL_Import_and_Export I think MeshLab is the reference program to check it. If it shows errors there, then something is suboptimal. However, that's quite common, I had this with many files before. It's maybe the exporter of a lot of 3D programs, idk. To use your design in OSCAD I can simply create a placeholder with circa the same size, so it should work. Btw, I did scale it at first but also tested it without scaling. >>18739 I was able to get rid of the error in the preview of the doll part by adding convexity=5. This whole exploration of the import function actually had a good side effect. I didn't know or forgot how OpenSCAD can create 3D objects from something like a matrix. I think I saw this before with someone creating a 3D map. But just looking at this yesterday I realized we could probably also use this for faces. >Currently only PNG images are supported. Alpha channel information of the image is ignored and the height for the pixel is determined by converting the color value to Grayscale using the linear luminance for the sRGB color space (Y = 0.2126R + 0.7152G + 0.0722B). The gray scale values are scaled to be in the range 0 to 100. https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/STL_Import_and_Export#Images
>>18683 I started to import different arms into OpenSCAD for comparison. So we can see what we can work with or alternatively what should be at least the reference. Rhetorical question: Which one is it gonna be? BJDoll or Dexarm from therobotstudio? I posted a comparison before, but deleted it, after realizing that it was unfair, since the BJDoll arm included the shoulder so the robot arm looked even stronger than it should. One basic problem remains anyways: We can't just scale it down while ignoring the parts that are supposed to be inside. So I guess, we'll (I) have to do it more or less from the scratch.
>>18780 I can create a cylinder with circa the same size of a upper arm in this case. Trimming it down at some point later of course, and use it as a reference for what would fit inside. In that case probably not that much is needed, depended on how much the lower arm is supposed to lift. The shown cylinder size works only for a hard-shelled design, of course. Personally I want to go for soft body waifus, but for my simple design such a hard shell is good enough for the now. The general idea here is, that I can use the same cylinder or a internal design based on that with any alternative arm shell, which might come with a different license or look different, whatever we want to have. But I'm not wasting my time with creating every outside shell part myself in OpenSCAD, especially not right now. For the same reason I'm also going to look around for an alternative upper body (chest). Some parts can be scaled and don't really need a lot of changing on the outside. Only the internals and connections might be done in OpenSCAD. I can also still design completely new parts at some point later.
>>18782 Premature size test with a Nema17 works, even with a 0.8 scaled down shoulder part (MaidCom chest).
>>18770 >But just looking at this yesterday I realized we could probably also use this for faces. That sounds really encouraging Anon. Good work. >>18783 That's coming together. My apologies if I missed it earlier, but is this a new face?
>>18783 Working with body parts is really more fun than figuring the details of some mechanism. I worked a bit on understanding the body size relations. I think it's a good idea in some cases to scale something down until it looks weird and then go back a bit above that border. Especially parts which are rather supposed to look cute and a bit fragile maybe. Her arm got way thinner than it started. As long as she doesn't look bulimic it's fine. For the thighs in a later iteration, I will still go with the grown up version of Alita size, though. I added therobotstudio dexarm as comparison into the second picrel. I really thought they would look good when I was watching his videos, but compared to something more human-like and petite it's over. He simply has a different use case. We have to go for the cute feminine looks first and then decide what we can do within those constraints.
>>18784 Thank you. Anyways, the parts are not from me, they're from that BJDoll I thought that I linked here: >>18710 (but didn't). From the last picrel there, here's the link: https://www.thingiverse.com/thing:2941323 Of course this comes with the downside that these parts might not come with the license you like to have ("created by Thingiverse user SorrowBJD, and is licensed under cc"). But aside from me being still undecided on this, I'm into working out how to have a framework where we can switch out parts anyways. My whole idea about using a cylinder to define the workspace here >>18782 was meant that way.
>>18790 OP if you're here, or if some Anon would care to bake, we need a new thread! :^)
NEW THREAD NEW THREAD NEW THREAD >>18800 >>18800 >>18800 >>18800 >>18800 NEW THREAD NEW THREAD NEW THREAD

Report/Delete/Moderation Forms
Delete
Report