/robowaifu/ - DIY Robot Wives

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

Roadmap: file restoration script within a few days, Final Solution alpha in a couple weeks.

Sorry for not being around for so long, will start getting back to it soon.

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB


(used to delete files and postings)

R&D General Robowaifu Technician 09/10/2019 (Tue) 06:58:26 No.83
This is a thread to discuss smaller waifu building problems, solutions, proposals and questions that don't warrant a thread. Keep it technical. I'll start.

Liquid battery and cooling in one
Having a single "artificial blood" system for liquid cooling and power storage would eliminate the need for a vulnerable solid state battery, eliminate the need for a separate cooling system, and solve the problem of extending those systems to extremities.
I have heard of flow batteries, you'd just need to use a pair of liquids that's safe enough and not too sensitive to changes in temperature.
This one looks like it fits the bill. The downside is that your waifu would essentially be running on herbicide. (though from what I gather, it's in soluble salt form and thus less dangerous than the usual variety)

How close are we to creating artificial muscles? And what's the second best option?
Muscles are perfect at what they do; they're powerful, compact, efficient, they carry their own weight, they aren't dependent on remote parts of the system, they can be controlled precisely, and they can perform many roles depending on their layout alone.
We could grow actual organic muscles for this purpose already but that's just fucking gross, and you'd need a lot of extra bloat to maintain them.
What we need are strands of whatever that can contract using electrical energy. Piezo does the trick at small scales, but would it be enough to match the real thing? There have been attempts, but nothing concrete so far.
What are some examples of technology that one could currently use instead?

High level and low level intelligence emulation
I've noticed a pattern in programs that emulate other computing hardware.
The first emulators that do the job at acceptable speeds are always the ones that use hacks and shortcuts to get the job done.
It comes down to a tradeoff. Analyzing and recompiling or reinterpreting the code itself on a more abstract level will introduce errors, but it is a magnitude of order more efficient than simulating every part of the circuitry down to each cycle. This is why a relatively high level emulator of a 6th gen video game console has close system requirements to a cycle-accurate emulator of the SNES.
Now, I want to present an analogy here. If training neural networks for every damn thing and trying to blindly replicate an organic system is akin to accurately emulating every logic gate in a circuit, what are some shortcuts we could take?
It is commonly repeated that a human brain has immense computing power, but this assumption is based just on the amount of neurons observed, and it's likely that most of them probably have nothing to do with intelligence or consciousness. If we trim those, the estimated computing power would drop to a more reasonable level. In addition, our computers just aren't built for doing things like neural systems do. They're better at some things, and worse at others. If we can do something in a digital way instead of trying to simulate an analog circuit doing the same thing, that's more computing power that we could save, possibly bridging the gap way earlier than we expected to.
The most obvious way to handle this would be doing as many mundane processing and hardware control tasks as possible in an optimized, digital way, and then using a GPU or another kind of circuit altogether to handle the magical "frontal lobe" part, so to speak.
>>5968 >but other than that it nails it and it's easy to tell when it messes up. Sounds great. I wonder how it can do such a great job at a task that surely involves a pretty deep understanding of the meaning of technical English words?
>>5971 People are just bad at writing abstracts and pad them with fluff. It mostly hones in on phrases like "we propose x", "x does y" and such. When it detects jabbering about something unrelated to what the authors did, it cuts it out of the sentence. Also AI is a lot more advanced than people realize. It achieves superhuman performance on some language tasks and outperforms average professionals in translation. The free machine translation people get at Google or Bing is dialed down a few orders of magnitude so the service doesn't bloat their server bills.
>>5980 >The free machine translation people get at Google or Bing is dialed down a few orders of magnitude so the service doesn't bloat their server bills. I see. Heh, I hadn't thought of that before, but it makes sense. So what you're really trying tell us is that anime will be real within our lifetimes?
>>5928 € 7,75 | laser radar 360 degree laser radar scanning distance measuring sensor diy wireless transmission infrared data transmission https://a.aliexpress.com/_mLplW5f It was even cheaper a while ago, but stated very clearly to be experimental. Didn't buy it, since I've already got to much to do and learn, also I don't want to put that into a regular robowaifu. Thought more of using it in some kind of Roomba or Narval-like floor cleaner.
>>6025 >also I don't want to put that into a regular robowaifu. I wonder if you can just set of video cameras at different angles in the rooms and then use those to tell your robowaifu where things are located/how they are moving?
Open file (269.59 KB 1228x393 Maixduino.png)
How do we avoid pozzed hardware in our robowaifus? https://www.youtube.com/watch?v=HNwWQ9zGT-8 RISC-V CPUs?
>>6338 We have an entire thread dedicated to this topic in general Anon. >>4506
I got some ideas for alternative energy sources. Sugar is a cheap source of energy that can be grown and turned into alcohol. How could we build an efficient generator that runs off alcohol? My first idea was to use a Stirling engine. Which gave me an idea it would also be possible to power it with oil. Sunflowers grow abundantly around here and roughly 10 sunflowers make about a litre of oil. With a conversion efficiency of 8% a litre would provide 0.8 kWh. Firewood could also be used if available. This would be pretty basic to set up and low risk. I already have a back-up power system and charging my batteries with something like this would be trivial. A generator could also be run by collecting rainwater in a tower and using gravity, or using mechanical advantage with a pulley system to lift a heavy object by hand to store its potential energy for turning a generator. A single person can only sustain an output of about 300 watts but it would be a way to store energy indefinitely. In good times a motor could be used to store a ton of energy in case of the sun being blocked out. It's also possible to run a hose from a stream for microhydro, but that's illegal. :^)
>>6420 >A single person can only sustain an output of about 300 watts Pftt. Plebeians :^) https://www.youtube.com/watch?v=1SUzcDUERLo
A Russian guy invented a compressor-free McKibben muscle using high-pressure electrolysis: https://www.youtube.com/watch?v=07oVoABYr10 The only issue is hydrogen gas is extremely explosive. :^)
>>6503 That guy's great, thanks Anon. I hope he becomes the IRL Russian Tony Stark tbh. :^)
New to this study, so bear with me. Has anyone experimented with the idea of 'visualization' in general AI? As in, you relate the words and definitions of a sentence to an image(s) to give it more 'understanding'/real world context of the concept itself, to better improve it's conversational abilities. For instance, if you were to say "I ate chicken for breakfast", it would look at all the verbs and nouns of the sentence and then look at the analyzed images of it that are coded to that word, then use the sentence structure of the language to connect those images in sequence. Think of it as the quick mental images you have when you read a book, but much lower-tech. Thinking with this it'd be good to go with a 'weak but smart' method, only knowing the 5,000 common use English words (at least at first), but developing a deeper, more human understanding of them, and it could be combined with normal NLP or have a custom system made. If this genuinely worked hopefully it could possibly be further developed later on hopefully to the point of creativity. Like creating a kind of artificial simple imagination, if that's even realistic. Do you think it'd have any kind of real impact on linguistic capabilities or would it just be useless? Is it nothing but fantasy? Is the technology there yet? I'd appreciate any kind of input on the subject, as I'm still learning.
Open file (67.42 KB 420x659 Leonardo_self.jpg)
>>6578 >New to this study, so bear with me. No worries Anon, just spit out whatever it is you're thinking of. We're a community here. >Has anyone experimented with the idea of 'visualization' in general AI? As in, you relate the words and definitions of a sentence to an image(s) to give it more 'understanding'/real world context of the concept itself, to better improve it's conversational abilities. For instance, if you were to say "I ate chicken for breakfast", it would look at all the verbs and nouns of the sentence and then look at the analyzed images of it that are coded to that word, then use the sentence structure of the language to connect those images in sequence. Think of it as the quick mental images you have when you read a book, but much lower-tech. Correlation between words and images has indeed been explored to some degree, both in the 'science' of Psychology, and the science and engineering of AI. And as you point out we all have the common experience of imagery within our imaginations that assists us in performing analogical reasoning--particularly when encountering something unexpected and new. So I definitely think you're on to something important there. OpenCV has some good tools for image analysis, and it wouldn't surprise me at all if an association-mapping system could be devised in just a couple thousand lines of compact C++ code using their API. >Thinking with this it'd be good to go with a 'weak but smart' method, only knowing the 5,000 common use English words (at least at first), but developing a deeper, more human understanding of them, and it could be combined with normal NLP or have a custom system made. If this genuinely worked hopefully it could possibly be further developed later on hopefully to the point of creativity. Like creating a kind of artificial simple imagination, if that's even realistic. I would argue we could probably get 90% of the way to our goal using just the most common 2'000 words of spoken English usage. Not only would this simplify any given contextual runtime analysis 'in the moment', but it would also greatly reduce the memory and other resource footprints needed for proper functionality. While I believe there are some things about the human mind irreproducible, I don't think that an imagination of sorts for our robowaifus is at all out of reach for us. >Do you think it'd have any kind of real impact on linguistic capabilities or would it just be useless? Is it nothing but fantasy? Is the technology there yet? I'd appreciate any kind of input on the subject, as I'm still learning. I'd expect that a functional, working system such as you're envisioning would indeed have a great impact on AI language generation and comprehension facilities. And I don't consider it fantastic, given our advances in both hardware and software capabilities. While practically all areas of AI that will do the 'heavy-lifting' in our robowaifus will take some clever engineering designs to shoehorn into small, mobile hardware systems, we'll manage it in the end Anon. And using creative, imaginative ideas like this is how we'll get there. Never be embarrassed or afraid to leap ahead with your imaginative forces Anon. We all have a great foundation of forebears' minds who have done the same, and they changed the world.
>>6578 There was a recent paper that combined pictures with word tokens and gained a significant improvement: https://arxiv.org/abs/2010.06775 It doesn't really form a picture or video representation of sentences though. It would be an interesting experiment to try. With a goal-oriented AI you could give it tasks to describe short video clips with a sentence and then also the reverse, generating clips from a sentence, using the same network that is doing chat and what not combined with a video encoder and decoder. Learning from artificial imagination has already been done with world models: https://worldmodels.github.io/
>>6420 Hi, I already mentioned "Direct Ethanol Fuel Cells" somewhere, probably in the thread for energy systems >>5080 or >>23 ... The output is acidious water and maybe some CO2. Burning oil isn't really interesting, since it would create toxic fumes and noise. Also think of all the moving parts, oil and weight. No thanks. Also, having waifus at home is the first step, and for that we won't need any of that. They can plug themselves in when sitting somewhere, or using wireless charging. >>6503 Yeah sure. Whats next, nuclear batteries? >>6578 We have a thread for chatbots and mind. >>22 I think, what you mentioned could best be done by graph databases and code to manipulate them. Thanks for reminding me, that I still didn't publish my code here, I was working on a while ago... I didn't really forget about it, just don't want to clean it up...
>>6845 >just don't want to clean it up... Then just put it here instead. As an R&D board, incomplete solutions are not only expected, but encouraged for quick information sharing and motivation/inspiration. Remember, we need to be quick on our feet here on this board!
Very interesting. So here's some ideas. 1. You could use Electropermanent magnets or motors and actuators. Found here. https://8kun.top/hover/res/48.html#q633 2. Motors could be put inside the bones to save space and hide them. Picture the motor inside the bone with a pulley that winds up something like fishing line. The line goes to the end of the bone around a pulley and then comes out in the same place that a normal tendon would attach. Line pulls to make muscle work. 3. There's several idea for origami type actuators in this thread. They could be placed inside the muscle and pulled by the string. This would also multiply the force of the motor pulling the string. 4. I really don't like the idea of pneumatics or hydraulics of any sort. I think in the long run they will cause a huge amount of trouble. Monitoring the force of these things and controlling the valves would be a nightmare. They would also be really noisy. If you going to have all these senor wires monitoring these things better to just go electric in the first place. 5. This idea may really bother some but I think it''s a really good one. No silicon skin. It's sticky and yucky. Why not use fabric? Look at the new artificial silk that they make T-shirts out of. This stuff is super slick, easy to keep clean. I believe some of it fairly stain resistant and doesn't grow bacteria on it easily which would be a good thing to have on a robot waifu. It could washed off in tub. Processor power gets good enough waifu will wash itself. Also we and everyone else will know the waifu is a robot. Instead of trying to make it look like human skin instead try for a human skin feel or even better than human which is much more important. Silicon doesn't come close. The fabric would also be warm to the touch becuse it would reflect your heat when you touched it. The effect could be enhanced by having a space blanket, Mylar coated with aluminum, right under the skin. 6. This brings me to the idea that all the muscles should be fabric and a pliant closed cell foam. Once again washable. The whole waifu could jump in the tub, scrub with a little laundry detergent then get out and dry. Foam origami could be structured like the origami muscles people posted earlier. A soft foam outside with a stiffer foam core that the string tied to a motor in the bone pulls to actuate muscles. Wrap the whole thing in fabric. This in turn is wrapped in a fabric skin made of fake silk. In order to keep it from sagging possibly some sort of Velcro could be used in strategic areas with rubber coating on the rest of the fabric skin and muscle to simulate the fascia that holds skin onto muscle in animals. The rubber would provide friction like the fascia but it would still move. 7. Since this thing is made of foam why not incorporate touch sensation into the foam. Seems I read that if you put carbon black into foam as it is pressed or squished the resistance gets less between the foam. Anyways this is a great thread and it got me to thinking about things could be done.
>>84 >"...What about internals? You might have to replace the inner lining of the mouth and uh, other human interface cavities once in a while. I don't have any ideas for those yet, perhaps something that binds when exposed to water, as opposed to the skin thing which would do better if it reacted to air. How do you refill liquids?...Self-cleaning is important as well, that's another use for water..." Another good reason to use fake silk fabric for skin. The private parts could be like a pair of shorts with internal pockets for private parts. The stomach skin could be pulled up and the abdominal muscles pushed aside and the internal pocket parts could be connected to a rigid internal keeper in the pelvis. The private parts could be surrounded by foam muscles providing different levels of tightness or even some sort of massage action. This also means you could have a hard plastic tube all the way from the mouth to the private parts and by drinking a lot of water wash itself out. Having a soft fabric skin that can be removed and replaced is huge advantage.
I need to make it a little more clear. Say you had a pair of shorts and split them up the sides. This is what I'm talking about but the private parts are built into the shorts. So the shorts are put on, the internal parts hooked up, then the shorts are tied or otherwise joined up on the sides and then the abdominal skin pulled down.
Another skin idea. Lycra Fabric? "...Lycra is a brand name for elastane, which is a highly elastic synthetic fabric. Despite having different names, Lycra, spandex, and elastane are all the same material, and these fabrics can stretch to 5-8 times their usual size... SO I measured my head and waist and if the only hole was in the head you could stretch, according to this, the whole body over the head. So it would be one continuous piece with a hole in the head that could be covered by hair. Velcro could hold the fabric on the inside of the lips with the private parts being pockets that are connected as you stretch it up the body.
>>6845 >I didn't really forget about it, just don't want to clean it up... Understandable, but in the meantime you're preventing potential inspiration others might get from your intermediate work. I'd recommend instead everyone simply post software as soon as it's reasonably functional, then provide incremental versions like the guy is doing in the C++ thread. Nothing will ever be perfect, software being one of the less likely areas. No one is surprised that things need attention afterwards.
>>8311 >>8312 >>8313 >>8314 These are some really interesting ideas Anon. Let me think about this for a while.
>>8311 >motors ... inside the bones to save space Yes. But the bones aren't big enough if the fembot is supposed to be as soft as a human. >Line pulls to make muscle work Yes, I had similar thoughts. But that's not enough. >origami type actuators Not sure if that would work (good enough) with motors. I'll look into it, to get an idea what you might have meant. >don't like the idea of pneumatics or hydraulics I don't see why this should be more trouble than motors, and why a combination would be bad. Also, there is no way around it, for partially soft robots. Otherwise the soft space can't be used. >No silicon skin. It's sticky and yucky. For all I know it depends how it is mixed. >Mylar coated with aluminum We'll see. I wan't my waifu to look human-like. Silicon can also be bend very often. It's gonna be a multi-layered skin anyways, if aluminium (foil?) would be good then it could be combined with silicone rubber as well. These fabrics with silver, against EMF make probably more sense. But I think more for sensors, I won't relay on body heat. Heating can come from the inside. > all the muscles should be fabric and a pliant closed cell foam Certainly not. You didn't do the dishes per hand in you life very much? Foam brakes down if it has been deformed often, it shrinks and deforms, it also holds bacteria, same for fabric. I think this would be okay for prototyping and cheap bots, though. But then again, rater with silicone rubber on the outside. I know about Lycra, Elastan, Spandex and Mylar for years. Had them in mind for exactly that purpose. Quite sure I mentioned that somewhere. It just about putting silicone rubber underneath or not, if not, then why? And why not putting some in between, letting it soak in? Or using the fabric only underneath for stabilization, but having silicone on top. It's futile to speculate about every detail here, just try it out when you have the body. Some details will then have influence on other decisions, in part it's about personal preference, complexity and price. >Having a soft fabric skin that can be removed and replaced is huge advantage. Yes, but it should also not move around unintentionally and it should have a lot of sensors in it, which would need to be connected to the body... Maybe you see the problem. There's more than one way to do it, and it depends. >>8316 >unfinished software release Unless it has bugs, breaks and is rather rudimentary...
>>8351 Yeah, just some confusion. I was quite sure that these power meshes or silk-like meshes had been mentioned quite often. It seems, not as often as I thought. The search results came from another anon, and where meant as help. My posting has been quoted, so the comment wasn't directed at you. Also, I didn't know you are new. Relax. While were at it, I wrote: > Foam brakes down if it has been deformed often, it shrinks and deforms, But didn't think of sofas and pillows, which also have some foam in them, which doesn't deform that much. So it might work somehow. It's just that using foam is also not a new idea, and whenever I hold a used sponge in my hand with that topic in mind, I'm getting skeptical about it. Till I realized that my couch is fine. >>8352 Why? It's not necessary to delete that conversation.
>>8354 >I was quite sure that these power meshes or silk-like meshes had been mentioned quite often. It seems, not as often as I thought. There are thousands of posts that hadn't been reposted here from the original board yet. Probably a tedious process. My guess is other posts about this will be in those.
>>8357 I had the same thought. However, it might be that I only looked at videos on Youtube about such meshes, downloaded them and thought about it. That we dicussed it here or back on 8chan more often, might be wrong memories. I never saw those as a alternative to silicone rubber, just as an addition. Such meshes would stilll need to be glued o the body or not? Then, how to prevent the silicone glue from soaking into the fabrics or the gaps in between? And what for? It is very flexible and can be washed, though I'm not sure about the details here.
Anyone having ideas about which bearing should be used in printed drives or in joints? For example, the diameter of stainless steel bearing balls. I was thinking of 2mm.
>>8364 That one guy's video showed him just using copper-clad BBs in his ball-bearing joint. Seems like a pretty cheap and readily-available supply approach to me Anon. >>8170
>>8365 Thanks for the reminder. In another video someone used the balls directly, so I want both, for trying out. However, I need to stop overthinking such shopping decisions. I simply bought a bunch of different bearings. Though, they're not copper-clad, for all I know. I watched this (commercial) here to get an overview: https://youtu.be/QhTI8CnRic8 - there's quite some variety. I'll have to look into that deeper, to understand when to use wich one, not only in drives but in joints as well.
Open file (65.26 KB 970x705 bearing-type.jpg)
Open file (81.94 KB 960x720 slide11.jpg)
Open file (146.74 KB 1280x720 maxresdefault.jpg)
>>8368 There are also quite some diagrams to find with image search on "bearing types"
>>8369 That's a lot of bearings Anon, thanks for the information.
>Is it possible to delete all my post here? How about we compromise and just delete the off-topic bits of the conversation, as well as the post that initially triggered your response? Our apologies to you for this misunderstanding Anon. We didn't consider the fact that a search posting in the same thread would create an inadvertent crosslink to your posting. You apparently took this to mean the search results posting was directed at you, which it wasn't. We'll try to keep search postings confined to the Library thread in the future to help keep this kind of thing to a minimum. Hopefully this clears things up for you. Let's just start over fresh here please. >>8314 >Lycra, spandex, and elastane are all the same material Here's a new (and consolidated) waifusearch, Anon: >>8431
Ok I've been looking around and there's mention of electrically actuated polymers here but it seems that the right term to search for data on it was not used. Although it may be one of those cases where it's rapidly changing. Here someone mentioned research on electically activated polymers https://alogs.theГунтretort.com/robowaifu/res/94.html#8468 After looking around a little I found this term and a wikipedia page https://en.wikipedia.org/wiki/Dielectric_elastomers This is a rapidly changing field. The Columbia university muscles are strong but very slow. Looking at other sources there does seem to be some that are strong and fast. The key point here is this is something that can be readily prototyped. All these motors, gears, etc. take a great deal of machinery to build but this stuff only takes polymers and electrodes. Way simple to experiment at low cost. I think has one to the best chances of being something done with low resources. So I found a book on these called "Dielectric Elastomers as Electromechanical Transducers: Fundamentals, Materials, Devices, Models and Applications of an Emerging Electroactive Polymer Technology" Here's a link to it. Hit one of the three links at the top to download. http://library.lol/main/6F7F8A9A06ABFA4C830FB7D6D8DB530D Here's videos. https://www.youtube.com/watch?v=PDqmGHHKkWw Really good video https://www.youtube.com/watch?v=LqOQQsig7og https://www.therobotreport.com/new-synthetic-muscle-step-forward-soft-robotics/ The key to these things is they are capacitors so all the efforts to raise a standard capacitors capacitance should be strategies to engineer these.
Dielectric Elastomers can also be used a sensors so the same tech can be used to sense touch. One bad thing about them is they don;t necessarily pull in so they would have to constrained or engineered to do so. Not sure how yet. Maybe a net of some sort like pneumatic muscles.
One thing you might could do is use aluminized mylar film or space blankets for electrodes and your polymer in thin films. The the only reason I'm so enamored with space blankets is they are so cheap. Like $2 for a 5' x 6' sheet. Roll the polymer and electrode around a rod. Make sure the roll can slide on the rod. Since the rod keeps it from expanding on the ends lengthwise when activated it will have to bulge like a blown up balloon to expand it's surface. A net around the blown up elastomer muscle will then pull to simulate muscle pull. Kind of screwy but it's a first thought.
A good page with links to 100 papers on Dielectric elastomers Open title in new tab then pick one of the links to download. http://libgen.rs/scimag/?q=Dielectric+elastomers
Thanks very much Anon. I really, really like the fact you are trying to concentrate on finding inexpensive/low resource methods for design and construction. This will help everyone out, even if they choose to use another approach. Having a wide array of robotics techniques available to us is good.
>https://www.youtube.com/watch?v=LqOQQsig7og If you don't watch this second video I linked then you are really missing out. In fact most of the whole entire solution is in that video. Not that the details are there but the foundation is all there. They are making computing circuits out of Dielectric elastomers. I mean they are making every single one of the elements needed for any computer out of these elements and embedding them "in" the muscles. What does this mean? By embedding computation you can multiplex instructions to the muscles and not have so many wires. You don't need valves or lots of motors with a huge number of interconnections. A thought. These elastomers could be printed in big thin sheets with a x-y plot table. With a few steps you could build up layers of logic and most important a lot of redundancy. They are using carbon for leads. Not so sure that will last. There's links here at this site for liquid metal leads. Might be worth looking in to but they would sure be more expensive but on the other hand if you multiplex the instructions to the muscles might not need that much in the way of leads. So you print out the muscle in big sheets then you just roll them up. Another idea is to have lines like fishing line under the sheets and print over the lines embedding them. When rolled up you have built in tendons attached throughout the muscle.
Now this is way the hell out there but it's not nuts. I also wish to emphasize that this is not easy and I'm only describing one path that could be taken that looks like the level of risk for success is likely to be very low. There's a thing called VHDL https://en.wikipedia.org/wiki/VHSIC_Hardware_Description_Language What it does is if you input a program of the "logic" of a computer in this language and if you have a graphical representation of how the computer chips elements are laid out for each of these "logical units" that is described in this language then the computer can lay out the logic elements to make a die to make the logical elements you have specified. Basically you decide on the computer logic and the software lays out the mask you would need to make it work. So once you define how these computing elements work in these elastomers then the computer could lay them out. Since we have such a big sheet of material for a muscle the elements could be large to keep the failures low. Computers could lay out drawing that could be placed over the thin sherets and you could squeegee on the circuits just like silk screen printing. In fact I bet many of the tools used for this could be repurposed. So all together lay out lines of strong fishing line for tendons, lay a thin sheet of elastomer. Use silk screening to add all the computing elements. Roll up the muscle and attach the fishing line tendons to bones and power.
Couple mistakes. The risk would seem to be low following this path. sherets=sheets Now one further thing. Since elastomers can be sensors, logic or muscle, we put sensors on the outside of our sheets. These senors are connected to the digital and analog converter and computer towards the middle. This also puts the tougher sensor part on the outside and the less tough logic on the inside. Since you have computing elements and gates you could have every centimeter a different touch sensor on the outside of the rolled up muscle and multiplex sensing them with the A to D converter. That data in turn sent to the processor to follow whatever logic called for. None of these elements need be super fast compared to humans since we are so slow compared to computers. Now for the muscles it would be too much trouble to have every single size and length so there's a way we could cheat with only a few sizes. Maybe only three or four. This would save us from having to make too many patterns for the muscles and computing elements in the muscles. Make the tendon lines very long. An example let's say we have 6 inch space but only have a 2 inch muscle. Just overlap several of them so you would have 3-2" muscles in a line then on top of that in the spaces between the three 2" muscles put two more 2" muscles. The tendon length would determine where they lay and each muscle tendon could be independent with it's own tendon, computer and power giving you great redundancy. In fact multiple muscles per muscle group would seem to be better. Combining them would the simple matter of lining up the muscles in their places and then gluing or clamping the various tendons in place. All of the individual tendons could be programed to act together. As for power some of these work on low power DC right now. A far out idea is to use millimeter wave length for power and communication to the muscles. Wave guides could be a simple metal tube 1mm or 2mm in diameter . The end of the tube could have a circuit to collect the power with a rectifier. The reason I bring this odd idea up is the new 5G is exactly that and they are planning a lot of cheap internet of things chips for these things. Certainly power receivers and transmitters will be part of this. The high end of this is around 300GHZ and at that frequency the size of a wave guide to send it with almost no loss would be 1 millimeter. So a 1mm plastic tube covered in aluminum foil or my favorite cheap thing mylar aluminized heat blankets could send power and could be used to multiplex data to the muscles. https://www.theunitconverter.com/gigahertz-to-wavelength-in-metres-conversion/300-gigahertz-to-wavelength-in-metres.html https://www.sciencedirect.com/topics/engineering/millimeter-wave I also wonder if you could not use some sort of salt water solution in conduits to carry power. This would carry away heat or circulate heat so the whole muscle would be warm like a human while doing double duty carrying electricity. Once again I'm not saying any of this would be easy and there's a lot of pitfalls that could happen but it does seem that this path would work with a lot of time put into it. If you could get the muscles right the computing power and AI stuff will be ready about the same time in 2025. You could have a fully operation waifu. And none of this would take gazillion dollar production equipment. Most would be repurposing stuff like silk screening, laser printers,, etc. Material cost would not be high. The mental effort would be high but a lot of it would be just making little parts and making sure you understood each little part then tie them together. I could not find a off hand free design for a microcontroller. The actual computing needs would be small for a muscle so it may be you could just plug away at it until you figure the base elements. You could probably use some of the ideas from RISC-V open source computer but it's 32 bit and probably way to sophisticated for our needs for a muscle controller.
>>8523 >I could not find a off hand free design for a microcontroller. The zipcpu may be in this computing class. I think it's an open design (not sure if any silicon has been forged yet though) >>4487 . Also, one of our own denizens is apparently working on his own project along this line >>7713 .
>>8529 >zipcpu Interesting. The cpu logic will definitely be a problem. People talk about using FPGA I wonder if these could be made in the elastomer and use the same tools? I assume FPGA would waste a lot of performance. It may be that we don't care just to move some muscles around. I'm really not qualified to answer that question. Basically I just happened to come along some of this stuff and while none of it is easy can you imagine tooling up enough motors or pneumatic vales combined with the mass of wires needed for sensors in a traditional hydraulic or pneumatic design? It would be a nightmare. The tooling involved to make it yourself would be financially crippling. However if you could get one good design for the logic and the muscles then elastomers allow repeating units that can be done in software and it would be likely you could silk screen this stuff like t-shirts. Cost plummets and you could get very elaborate in features as long as you were willing to put in the work to make it happen. Also collaboration could happen because you could get by with a few elastomer logic, sensor and computing elements. All the rest being software. All the work could be parsed out over time in little pieces.
Since these elastomer muscles could be embedded you could use inch worm type pumps to have a circulatory system. It also could be that elastomer pumps could be used like mini-hydraulic systems to amplify force. Some of the very strong elastomer motors are real slow and some are real fast. It may be the fast ones lack power(not sure). By using the speed to drive localized pumps you could increase force using the same type muscle all over. Once again we limit cost by reusing the same basic elements over and over.
This set of programs by DARPA would seem to be relevant. It's a older program not sure how far they got but it's all about automating the hardware if you know the elements involved. It's for silicon but I suppose you could add in the parameters of elastomers and then use the same design tools. They focus on allowing a small group to engineer a large system on chip design with three or four people as compared to hundreds. https://spectrum.ieee.org/tech-talk/computing/hardware/darpas-planning-a-major-remake-of-us-electronics-pay-attention
Not sure if we can use that for anything, but it's fascinating: https://youtu.be/LU77kPf25Yg - a expanding pulley. He also mentioned a old book, where he has it from. It's free: https://books.google.es/books?id=MWlRAQAAMAAJ&newbks=0&redir_esc=y (pdf uploaded, epub format not bc server doesn't allo it) He says it's basically a cam and roller mechanism, which converts rotary motion to linear motion, but a special one. These mechanisms are also interesting by themselves, and its good to know about them. Also, this window mechanism might be usefull: https://youtu.be/IgZDlnpTFVw - or something similar inspired by it (muscle? rolling it up?)
Open file (45.59 KB 311x191 Selection_276.png)
>>8728 Interesting book, thanks. I could see this double-wishbone universal joint being useful for transferring rotary motion down a limb across a joint boundary. For example rotating a wrist-joint remotely, across the elbow-joint boundary. > p26
Open file (81.96 KB 236x184 ClipboardImage.png)
>>8728 BTW, I just had a humorous idea from this design notion. If we're going to be able to rotate a wrist via a drive rod in this fashion, who say we have to stop with the rotational limits of the normal human arm? >eg >>7719 >Tibia and Fibula Why can't we have unlimited rotation of the wrist? Just lock the grip off (onto a handle, say) and then just start rotating like a slow-turn drill (just like in my space movies). Ofc this would only work for fully self-contained hand actuators, but the idea is pretty interesting.
>>8730 Depends on how human-like she is supposed to be. Won't work with seamless skin. Also, this would require that one thing I want to try out is going to work: Not using cables to connect the limbs, but creating bones that can run current and always touch each other, but also won't short out. Information would then need to be transmitted via light signals, from limb to limb.
>>8732 >Won't work with seamless skin. Agreed. But one thing most of us here seem to approve of generally is not trying to be too realistic with our robowaifu's designs at this stage. cf, the idea of serviceable 'eye assemblies' that Anons discussed. >>4355 You can even weave the look aesthetic into the waifu's design itself, kind of like was discussed about Sophie-bot. >>8392 >"...maybe a couple robowaifu-aesthetic half cups..."

Report/Delete/Moderation Forms

Captcha (required for reports)

no cookies?