I was looking for the link on using optical fibers linked to camera chips for sensors (this is one of the greatest ideas in history). The link for using cameras and fiber for sensors is,
https://hackaday.com/2019/08/30/fibergrid-an-inexpensive-optical-sensor-framework/
While looking for this I found a new link that really fits in well with the other that I had not seen. This one,
https://hackaday.com/2011/10/21/building-optical-flex-sensors/
A little thought and you will see this guy has figured out a way to make very good touch/pressure sensors that could be waterproof, reliable, self contained and if you combine this with the first, cameras used to sense the values of light back from sensors, now you have a complete sensor package. You can use optical encoders fed by fiber(light) for motion and use this second idea for pressure, touch, sensing and with the first, cameras for sensing, you have the whole thing in one nice package. If the touch sensors are in a X-Y grid you have position sensing, which could be combined with the light attenuation to give touch pressure values with point locality. Since cameras these days have huge values you could have a very precise position location sensing.
And the whole thing in a waifu could be self calibrating. So you plug all these fibers and/or sensors into a package. The waifu bends it's limbs in sequence under control while monitoring the feedback from the sensors. After it calibrated what joints go to which pixels on the camera it starts squeezing it's body parts and calibrates the amount of squeeze with pressure and calibrates which pixels and by how much they change. The whole done in a few minutes. Could even do so periodically to make sure calibration is correct with wear.
Now the trick is to get away from fiber. Stringing together all these fibers is labor intensive. The second link shows a path to this. In it he has a hollow tube with a filler. What if you made a trough for one side of a skin section and on the other side a hill that fits into the trough? The outsides would be a dark rubber. So you flex it and the light hits the dark rubber and is absorbed as opposed to going straight through the thin film which is clear that you made the trough and hill out of. When I say film I'm saying you make this structure, like a big flexible nervous system out of a cast plastic. Possibly cast the clear interior then spay over, paint or cast the exterior dark surface. I suspect that there will be some trickery and art involved in making this film, trough and hill structure to get the maximum sensitivity without blocking the light all together.
You would have to separate the joint bending light attenuation from the pressure/touch sensing light attenuation. I suppose you could bend, then pressure test, touch calibration. Since it knows it's bending joints and knows this value then any further attenuation of light would be assumed to be touch.
Every since I heard about the camera fiber sensor idea I've been trying to think of a way to make this THE method of sensing for all facets of the robowaifu I think, "this is the way", though I have no doubt there are lots of niggling bugaboos that could cause trouble. But once you could get some small section worked out and function properly then it's only a matter of multiplying that small section onto the whole of the waifu.
Message too long. Click
here
to view full text.