/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


Who wouldn't hug a kiwi. Robowaifu Technician 09/11/2019 (Wed) 01:58:04 No.104
Important good news for Kiwi! (>>14757) === Howdy, I'm planning on manufacturing and selling companion kiwi robot girls. Why kiwi girls? Robot arms are hard. What would you prefer to hug, a 1 m or 1.5 m robot? Blue and white or red and black? She'll be ultra light around 10 to 15 kg Max with self balancing functionality. Cost is dependent on size, 1000 for 1 m or 1500 for 1.5 m. I'm but one waifugineer, so I'm going to setup around one model to self produce. If things go well, costs can go down and more models can be made. Hopefully with arms someday. >=== -add news crosslink
Edited last time by Chobitsu on 12/23/2021 (Thu) 06:43:53.
>>6175 I don't know about the weird nodule on it's head, but with a little more plushie and a little less exposed plastic this looks fairly close to what some anons have said they'd be happy with tbh.
>>6176 The weird nodule on the head is the 360 degree LIDAR. I've been looking for a nice way to have 360 degree vision for my waifu and having a weird nodule on her head suddenly looks like the best cute compromise.
>>6177 Integrate the sensors into a hat.
>>6178 Or a big crystal like that female sensei character, sorry I don't know her name but I've seen her in images beside Ellen Baker.
>>6179 Actually, it just dawned on me that one of our earlier threads has a crystal-hat waifu, Aoki Lapis. ]]73
>>6175 tbh these things appear to be A) brown, and B) wearing hajibs. trying to brainwash little nips into how awesome miscegenation with mudslimes is? i get that these are simply toys, 'so what?", but i've become extremely distrustful of jewish globalism agendas. i prefer to see the Japanese remain Japanese, not just another EU-alike cesspool.
>>6180 I was originally going to say you read way much into things, but then this is from one of the Pepper creators, so as mainstream as robots get, so I dunno. The bigger crime to me is that this thing costs 3000 bucks for one piece or 5K for a two-pack. Cute but otherwise useless robots should be in 200-500 range.
>>6182 whelp, meant to answer >>6181
>>6182 >>6183 A) you've probably forgotten what it's like to be a child, and how gullible they are, and B) Never underestimate the sinister designs of Evil and it's puppets, evil men and women.
>>6136 > I'm not even bothering to fit any electronics inside my robowaifu aside from servo motors which drive the joints. There are so many structural and movement concerns that at the moment I am keeping the electronics separate and joined to the body with long wires. Once I have the structure and movement figured out, I'll maybe add mounting points anywhere there's space to conceal a circuit board or two, but I can see my robowaifu being connected to a big box of tricks at the moment.
Open file (36.27 KB 173x253 1619851210446wew.jpg)
this thread is kinda dead but I'd recommend using your time for voice recognition and endearing expressiveness, it'd be a lot simpler and less time consuming/realistic than getting a good voice output program going, and it being bad could even make it worse so maybe some little simple wings, some legs, and a cute face, with voice recognition and facial/body expressions, but mute that functionality will of course be added later, but much more important than functionality upon base release is the initial push into more people's radar to bring in more money, innovation, and competition, kind like with VR headsets
>>10227 Good points Anon.
>>10227 (Part One) To further elaborate on this idea, and then the technicalities of it and its relative benefits to a completely human body, take a look at this harpy (image one, very very cute, artist is rnd.jpg) and imagine that you want to build something along the lines of this model. The biggest and most apparent differences are, of course, the wings and the legs, which you may think would complicate the build further, but it's the contrary, as it would eliminate; 1. The needs for human hands and feet, which have just a fuckin shitload of bones in them, which in turn makes for much more complex construction. for instance, a single bird's wing (image two) only have three joints from the main connector at the 'shoulder', compared to humans, who have 27 in just a single hand, a much smaller and dense space. The legs (image three) have even less, with a single 'ankle' joint and a 'knee' joint, although depending on the style of feet for the bird species the toe phalanges (up to five a toe in some species!) may give some trouble, but that could be circumvented by simply creating the foot of a species with less of them lol So not only would the construction be much easy, but it would have a higher relative quality. 2. The problem of the uncanny valley that could be triggered by limbs and/or appendages trying to perfectly resemble those of humans and failing. We have such a response to the uncanny valley because we're socialized with humans to notice even the slightest thing that's off, but we don't have nearly the same capability with other species, on top of the anatomy of certain ones being much easier to deal with. Both those points are pretty applicaple to several different species of monster girls as a whole, and make them, in my opinion, a much more viable alternative in this stage of robowaifus, much in the same way some anons push for RWs that go out of their way to not appear human while still being endearing. (shoutout to /monster/) (Part Two) So on the other side of this proposal is the actual interaction and how such things could be done for max 'oh god oh fuck I wanna cuddle you'. Both for capitalistic budgeting/marketing and also because I would want one for myself. So essentially it would be a mute, as I've said in my previous post, because frankly the technology for irl, face to face conversations isn't there yet, and probably won't be for quite a while. So all that time that would be otherwise spent on the 120(!) joints in the hands and feet alone and the voice output system could instead be pumped into two main interaction systems that feed into each other for a wholly more immersive waifu experience. Firstly, the voice recognition system. For just a basic setup, let's say that, before your customized harpy petwife shipped out, it has both it's own name (picked by the owner) and the owner's name/face already calibrated into it. It would use a kind of emotional recognition system via voice (and likely a small visual system as support) to know when to perform various actions. Say, you're feeling kind of down, it'll come waddle over and cuddle up to you, etc, etc And this is where the expressions start to come in. So secondly it would have a complex series of body language and facial language, paired with a haptic systems. Say if you were to pat its head, it would have give a more content expression, but if you were to suddenly stop after a very small amount of time it may try to cozy up and look up expectantly to you in an attempt to get more. Or, alternatively, if you don't pay it enough attention, it may become dejected (you monster) and ruffles its feathers at you, which reinforces your desire to take care of it and incentivizes daily bonding. If you hold it in a standing position long enough, it may wrap its wings around you, or if you hold around it in a vertical/sleeping position long enough it goes into sleep mode. etc, etc. And you can fuck it, of course. I've got a couple ideas for aiding with the facial emotiveness, which is the most important thing in RWs, most notably shifting eye colors depending on the 'mood' wheel (corresponding with the color wheel), but I really don't know how else to go about it for the anime aesthetic, and it seems neither does anyone else, barring just a screen for a face. The Mirai head (image four) from Furhat Robotics/Bandai Namco is a step in the right direction, but I'm still kinda worried that this aspect is lagging. In closing, again, this is more just a baby step towards a total interactive relationship with full voice capabilities, but it's more important to establish the market for robot companions (not even necessarily in a romantic sense, although that's most likely how it'd turn out, but in a pair bonding sense) as a way to bring in investors, and therefore competition, and therefore innovation, resulting in more and more advanced upgrades to your once-mute bird wife until she can somehow make you miso soup with her bird... hands... every morning and reaffirming her love for you verbally I went in pretty deep, probably should've just made a thread about monster girl robowaifus, but I didn't want to bump one off the board (image five: endgame)
>>10237 Very nice post Anon. I liked the details you went into about the bones/joints, and the /comfy/ homelife scenario you painted of Anon and his RoboPetgirl . Just like as seems very commonplace in my favorite Chinese Cartoons on the subject, and as you brought up too, some way to solidly imprint a waifu (whether RW or Petgirl) onto her master alone needs to be solved. The obvious idea is some type of bioinformatics scanning system for the Anon to use. I'd be hesitant to use just one method -- say voice -- alone. Too easy to mimic. In general I'd say emotiveness is a very fundamental part of Character Appeal (one of Frank & Ollie's 12 Principles) , and is something we all need to be creative about addressing well. Also, the eye color-shift thing is a nice idea.
>>10237 Glad I found your post anon. A sub one meter waddling very fluffy harpy petwife that makes cute squeaks sounds really nice. Will save your "features list" for reference. As for recognizing her master, we can start really simple and have an RF (radio frequency tag) ring which the owner wears, and the robot will just triangulate the location of that ring. Then only voice or facial recognition if the system is secure and not in the wrong hands.
Open file (200.98 KB 1920x1080 DenkiShoujo.jpg)
Open file (22.53 KB 245x440 Femisapien1.jpg)
Surprise! I graduated Summa Cum Laude and lost my job. Now I want to dedicate myself to building a robotics company for robowaifus. For now I need to come up with something that can be rapidly fabricated and sold for a decent profit. I'm considering making basic miniature robot kits shaped like battery girls from Rumble Garanndoll. Though, I'm very open to suggestions. Maybe a simplified Femisapien inspired kit?
>>14757 >Surprise! I graduated Summa Cum Laude Amazing job! Earnest (and well-merited) congratulations Kiwi!! I know you put an amazing amount of work into that achievement. >and lost my job Well, they say that the Lord doesn't close one door without opening another one! :^) >Now I want to dedicate myself to building a robotics company for robowaifus. Ebin news Anon. I wish you great success, and I'd like to help out in some way if I can. Please let me know if there's anything we can do to help you Kiwi. >For now I need to come up with something that can be rapidly fabricated and sold for a decent profit. I wish I had a complete software package ready to freely give to you. Unfortunately that's a good ways off still. You are exactly the target 'market' for my works however, and why everything I do here is MIT-licensed. >Though, I'm very open to suggestions. I don't feel particularly qualified to make any myself personally--but why not your very namesake idea ITT? It was a good idea all the way back then, and it's still a good one today. I bet thousands of anons would pay a little to be able to have a cute little Kiwi daki that they could snuggle, and that would talk to them. You could probably expand that into a line of child's plushies that could conceivably have tens of thousands of sales. The chat bit it probably key, so you may need to invest lots of groundwork into finding a reasonable solution for that. Certainly it would also help Anons on /robowaifu/ as well to have a solution to that, myself included. Again, great congratulations Kiwi. You deserve a big celebration soon! Please let me know if there's something I can do for you Anon. Cheers.
>>10237 As a fellow harpy chad, I think you're onto something. I do not believe in robots having bio inspired bones. I took too many A.P. biology classes to want that craziness. But, there are some aspects of harpy anatomy worth copying like the structure of the tendons and muscles. I completely agree with your assessment of the uncanny valley. Monster girls like harpies are so different they should bypass the feelings of unease. Would also help her stand out from accusations of selling sex dolls. >Max out 'oh god oh fuck I wanna cuddle you'. Now that is a good design goal. Using sensor fusion to obtain contextual activation cues for behaviors is actually pretty clever. Bonding and you can snu snu her will help with sales. I've been considering lithophanes and touch screens for faces but both aren't great. We really need a way to properly translate anime girl faces into 3D. Feel free to join me, if this leads to a successful company then you'd be hired. >>14761 >Just like, sell waifus Anon Yes, I will Chobitsu, I have failed trying to make her countless times but, you are right. The only way to go is forward to cute monster girls. So, I'm thinking of a design that's a cross between Toki from Kemono Friends, Kikimora, and Roll. (Roll objectively has an amazing design for a robot, well balanced with high ground contact area.)
Open file (137.80 KB 705x786 CurvedKiwi.jpg)
>>14768 Tried a curved color lithophane and she feels cute and 3D in person. Until a viable alternative is found, this is the path forward.
Will be creating a new thread once a new circuit, KiwiMeido, custom servos, and various other tidbits are designed. Lots of failures will make way for something eventually.
>>14768 >Yes, I will Chobitsu, I have failed trying to make her countless times but, you are right. That's the spirit Kiwi! :^) >The only way to go is forward to cute monster girls. Lol >So, I'm thinking of a design that's a cross between Toki from Kemono Friends, Kikimora, and Roll. (Roll objectively has an amazing design for a robot, well balanced with high ground contact area.) I like Roll's design too.
>>14857 >Will be creating a new thread once a new circuit, KiwiMeido, custom servos, and various other tidbits are designed. Lots of failures will make way for something eventually. Looking forward to it. I wish you great success in your Robowaifu business Kiwi.
>>14933 >>14934 Small update, decided on a final battery charging circuit.
>>14993 It's designed to be reflow soldered with SMD components. This cuts down on costs/size/mass and SMD reflow soldering is the easiest. My intention is to create an open source design and bulk order them to sell at near cost so it'll be the lowest cost/most practical option for fellow Anons. On that note, I'm still conflicted on if I want to integrate everything on one board to simplify the design and lower overall cost or, go for a modular design to enable better customization at the cost of well, cost. I welcome feedback from all Anons in the matter. >Bad professors Been there AllieDev, had to retake a machinist course because my professor sucked. Highly recommend going to a different college if you do intend to go back. Do what will lead to the best life for you.
>>14995 SMD reflow soldering I had completely forgotten this process exists! This is probably how my servo circuits were made (place hundreds in an oven and reflow solder them all at once). Although I noted that the hobbyist ovens often leave some manual touching up to be done. I was scuppered when attempting to repair SMD components by hand. I must have had my soldering iron too hot because even though my dwell time was overall just a few seconds, I still managed to lift off the circuit board substrate and destroy the tiny solder pads. Left a nasty tear and scorch marks. I also probably should have made better use of my desoldering braid rather than touching the tip of the soldering iron directly to the solder pads. But it is SUPER fiddly. Seems everyone I see manually soldering SMD components on YouTube uses a freaking optical microscope! (Though this could be just for recording purposes).
>>14989 >Small update, decided on a final battery charging circuit. Splendid work Kiwi. Your efforts have made us move ahead in beginning to think through some power/battery/charging topics in advance. Keep it up! >>14995 >On that note, I'm still conflicted on if I want to integrate everything on one board to simplify the design and lower overall cost or, go for a modular design to enable better customization at the cost of well, cost. I welcome feedback from all Anons in the matter Cost is a primary constraint for practically any man wanting to create a DIY Robot Wife in his bedroom/garage lab. I would suggest that for the time being. Once your business venture is established and in-flight, then more developer-focused efficiencies for customizations will be a natural flow to the 3rd-party robowaifu aftermarkets? >>15010 You are very likely our most experienced Anon for this specific topic of soldering Anon. We'd all welcome more detailed inputs from you going forward.
Open file (2.09 MB 2437x3110 20220215_144754.jpg)
Open file (4.93 MB 4128x3096 20220215_144919.jpg)
Open file (4.49 MB 4128x3096 20220215_144932.jpg)
Open file (5.34 MB 4128x3096 20220215_145001.jpg)
After much effort and looking at the works of other Anons, I've come up with a super light leg that is durable yet slightly yielding due to the inclusion of wooden rods. Wood is the future of waifu's as far as I can see because it's cheap, light, and easy to work with.
>>15214 Brilliant work Anon! Elegance in simplicity. I appreciate the geometric layout you've chosen. Obviously meticulous attention to form-follows-function. If it's not burdensome design-wise, I'd request you to look into creating a proper 'double-jointed' knee for this design. It would be more accurate biologically, and would more properly deform an outer, flexible overshell. Glad to hear from you Kiwi. Nice progress! :^)
>>15222 Incorporating a double jointed knee would be difficult but, I'll try it
>>15226 I'm a bit out of the loop here, would the double jointed knees give any advantage over having a single jointed knee? Just seems like it would make the leg much heavier since you'll now need an extra servo on each leg to move the second knee.
>>15249 The advantage is the leg will have a kneecap and move in a slightly more human-like way. I refuse to use a redundant servo as it would add mass and consume power for no reason. I'm trying to develop one that'll work passively or powered. Turbo_SunShine's PCB holder is currently my inspiration. Who is she? She's so cute!
Open file (134.25 KB 1000x1000 g7b2eghj12nvb.jpg)
Open file (3.81 MB 4000x3000 82b17tegy1u2v.jpg)
Open file (418.08 KB 1080x852 basemesh.png)
>>15214 I don't like this insanely super lightweight approach you keep bringing up. Light materials do little good when so little as the breeze of a door or walking past them is enough to knock them over. They are best for cosmetics, but not so much for the base frame imo. You also inadvertently made a mini easel you just put on a scale and took several pictures of. lol. Had made some progress through the basemesh of my custom Pandora. About 1/3rd of the way done.
>>15253 I understand what you're saying, I am going a tad overboard with my fixation on making her have as little mass as feasible. Though, reploid boots would keep her from falling over either way. I do see the similarity to an easel lol Pandora is looking good AllieDev, I have high hopes in her.
Open file (1.31 MB 1500x2037 78348131_p0.png)
Open file (32.80 KB 520x350 birdbones1r.jpg)
Open file (105.74 KB 1440x1440 kiwi skeleton.jpg)
>>15250 >Who is she? She's so cute! I think she's called Diva Media from World Flippers. >The advantage is the leg will have a kneecap and move in a slightly more human-like way. I see, so you're using double jointed knees to be able to make a more human-like kneecap? I thought you wanted to make 2 knees on each leg just like how birds have 2 "knees" Your design is interesting but why didn't you choose a more straight-forward knee design?
>>15260 >Dive Media I love her and her bizarre pinball RPG. >Human-like kneecap That is correct, though your misconception of me building an actual robot kiwi is great. The knee was designed so that she could sit seiza, in a chair, and stand up straight with one servo per hip to actuate her knees. Doing so is more difficult with a double knee. For what it's worth, I made this thread because I was deeply depressed and lost in despair and this stupid cute Kiwi Kemono Friend kept me from acting stupid again. Working on her gives me hope.
>>15261 when are you going to share any of that skeleton you've been working on for the last few months?
Open file (109.24 KB 850x1137 DodoCute.jpg)
>>15266 The leg was my first real success. Made a chest and waist but they were either too heavy or too weak depending on infill and foam filling causes things to balloon. Basically what MeidoDev made but worse. Right now I'm redesigning everything based on 12 inch by 1/4 inch wooden rods, which are pretty strong, cheap, and easily accessible. I try not to post as many failures but if you'd prefer it, I could posy more failures.
>>15226 I really think it will be worth the efforts, Kiwi. Godspeed. >>15249 A cute! >>15253 I like where you're going with Pandora, AllieDev. Clean lines tbh. >>15254 >I am going a tad overboard with my fixation on making her have as little mass as feasible. Actually, IMO you're clearly a sound Mechanical Engineer. The lowest mass feasible (in nearly every component of our robowaifus) that doesn't compromise structural integrity or violate other important principles is only to our benefit as a group overall. >>15261 >For what it's worth, I made this thread because I was deeply depressed and lost in despair and this stupid cute Kiwi Kemono Friend kept me from acting stupid again. NUUUU! She's not a stoop :^) >Working on her gives me hope. Then she's all the more dear to us Kiwi! Please make her real Anon. >>15268 We all can learn from each other's mistakes right? Just look at me, week after week and I still don't have it right! :^) In fact, it'll clearly be years tbh
Open file (82.96 KB 850x1297 KiwiTan.jpg)
>>15296 It's a tad frustrating to realize it will take years but, all that matters is that we get there. The end point of cheap yet capable waifus for all is worth everything. Kiwi will be made, in time.
>>15308 It feels like I am giving Pandora a second chance after doing some research on her backstory. So I understand the feeling.
>>15316 She'll be quite lovely but she will require patience and persistent focused effort. Keep the faith my friend, we will reach her.
Open file (4.92 MB 4128x3096 20220301_145434.jpg)
Open file (1.21 MB 3759x888 20220301_145330.jpg)
Made a new leg, double knees are too complicated. So now she has the same mechanism as before, just significantly streamlined with a pronounced knee that should deform silicone in a manner similar to how a knee cap deforms skin based on anatomical analysis. Taking inspiration from RiCODev, I'm only making minimal changes then moving on. Perfection is the enemy of completion.
>>15366 Thanks for the beyond-the-call efforts Kiwi! :^) I'm sure she will be fine that way, and it sounds like it was an opportunity to create a more-streamlined design too? I believe you are on exactly the right path for us to quickly achieve near life-sized robowaifus for Anons with modest resources. This is an exciting direction actually, please keep moving forward!
>>15366 >Perfection is the enemy of completion. What? These pictures don't really illustrate the elbow joint very well.
Open file (49.51 KB 510x386 AscentoLegMechanism.png)
>>15374 Completion requires an end point. Perfection can be said to have no end as improvements can be made ad infinitum to fit various ideals. I want to restart again to inch closer to my perfect leg but, that's a potentially endless cycle. Focusing on getting something done is better. The knee functions nearly identically to Ascento's leg, just mass optimized and legally distinct. Not pictured is the use of rubber bands for gravity compensation as a lower cost alternative to Ascento's springs.
>>15414 What a cool design Kiwi, thanks.
>>15414 Isn't it a problem to have a sping, or even worse a rubber band, being loaded all the time if the joints are in a non-neutral position? I like the idea with the spring, not I'm thinking if I could use that as well, but maybe with a motor controlling the non-neutral position of the spring.
Open file (102.78 KB 800x600 019.jpg)
>>15746 Your suspicion is understandable. Springs and rubber bands are meant to experience stressed states for prolonged time. It is true that some springs will suffer from fatigue under prolonged or extreme stress. All you need to do is ensure your design does not put too much strain on the elastic element. For us, rubber bands are easy to design around and are used frequentally in DIY RC cars and robotics. (Picrel is a lego RC car which uses rubber bands for power transmission and suspension, used for the sake of clarity, note that all rubber bands are in a constant state of stress.)
>>15748 That's kinda cool looking Kiwi. I think the wide variety of elastic bands broadly available, and their low-cost in general make them a natural fit for our goals here on /robowaifu/.

Report/Delete/Moderation Forms
Delete
Report