/robowaifu/ - DIY Robot Wives

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

LynxChan updated to 2.5.7, let me know whether there are any issues (admin at j dot w).


Reports of my death have been greatly overestimiste.

Still trying to get done with some IRL work, but should be able to update some stuff soon.

#WEALWAYSWIN

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)


Welcome to /robowaifu/, the exotic AI tavern where intrepid adventurers gather to swap loot & old war stories...


Open file (32.62 KB 341x512 unnamed.jpg)
Biological synthetic brains for robowaifus? Robowaifu Technician 04/06/2020 (Mon) 20:16:19 No.2184
Scientists made a neural network from rat neurons that could fly a fighter jet in a simulator and control a small robot. I think that lab grown biological components would be a great way to go for some robowaifu systems. It could also make it feel more real. https://www.google.com/amp/s/singularityhub.com/2010/10/06/videos-of-robot-controlled-by-rat-brain-amazing-technology-still-moving-forward/amp/
>>2184 It would be amazing and a real breakthrough if it could be managed. We've had similar discussions on this topic in the past and the general consensus is that adding biological components into a robowaifu drastically increases the complexity needed to support it, in addition to the many and varied other needs just to create a basic robowaifu chassis, control software, &tc., &tc. I personally think having it onboard is entirely out for these reasons, but possibly we could manage a bench life-support model some day. Either way it would be awesome to manage, and would dramatically increase both the cost and the complexity. I don't really see it being in the domain of a /robowaifu/-tier (DIY Robotic Wives) approach within decades. Still, thanks for the link and the new thread OP. I'd like to continue discussing it and see you all prove me wrong! :^)
>>2188 Someone on youtube tried to make one himself but failed. If it could be made to work reliably it could be far more energy and space efficient than simulated neurons. A human brain is no bigger than a grapefruit but you would need a computer the size of a small town to simulate it. We should at least try to make small scale ones so we can understand it well enough to scale it up
>>2199 >it could be far more energy and space efficient than simulated neurons Sure the tissue itself is vastly superior as a reactive computing fabric than anything man can devise (at least currently). However the life-support need seems to be a real showstopper afaict Anon.
Even people with more resources struggle to keep these cell cultures alive. I wouldn't want a robowaifu with an expiration date shorter than my own.
What if a way could be devised to quickly program a tissue biopsy? If having a small, additional neural system like rat brain onboard, and if it could be swapped out for new tissue when it expired and be quickly brought up to speed, would that suffice Anons? Just raise a bunch of rats and cut out a new brain every month say, to give the robowaifu very lifelike and coordinated motions sufficient to dance elegantly say, or to talk very believably. Would you be willing to do it?
>>2203 No, I wouldn't. My gut tells me this is would be an inefficient, error-prone approach. Biology inspired solutions are more appealing to me than actual biological parts. https://rebootingcomputing.ieee.org/images/files/pdf/4-rcs2-hylton_-_intro_to_neuromorphic_computing.pdf https://en.wikichip.org/wiki/intel/loihi
>>2203 Actually, as I think it through, you'd want two tissue cartidge carriers, and you'd replace the oldest for the new every half-life of the tissue sample whatever that turns out to be. the first half of the 'life-span' of the sample would be spent getting programmed to match the other sample, so that the neural pathways were similar before the older of the pair expires. lather, rinse, repeat. >>2204 fair enough. this is obviously the wild-frontier of biological research and in fact has bigger implications that for just robowaifus tbh. Thanks again for the research links anon.
>>2205 While this was ten years ago, these cultures could support 100,000 neurons(not all of which were functional). An actual rat has around 100,000,000. So that, shelf-life, and the issue of getting it to work are all obstacles.
>>2206 So, you're aware of Elon Musk's recent startup Neuralink, Anon? we highlighted it here on /robowaifu/ the day of the reveal One big advantage ratbrain has over human brains is you don't have to be careful about it. A sophisticated matrix of at least 100'000 gold microfine probes should be doable in a 1cm cubic volume today. And for a shitload less money than the $1Bn robo-surgeon setup they're using now. And, as Neuralink has already demonstrated, this is a two-way setup. The tissue can be 'written' to as well as read from. Ergo, the tissue can be quickly reprogrammed to no longer be a rat but instead become a match of the other tissue cartridge containing the ever-improving robowaifu personality & behaviors you've invested the training time into.
>>2208 I've heard of it, but I never thought of it being used in this context. Has it ever been used on cell-cultures? The harvesting, growing and maintenance aspect are what make it seem unfeasible to me. That and you would need a lot of them. Neurallink can't make a rat that can read books.
>>2208 probably should have 3 or even 4 cartridges active at any given time to help with failover capacity in case of a mishap of some sort. Also, it would be wise to have some way to dump the neural net out into digital form and backing it up in case of a complete failure due to disease, some systemic fubar, or zogbots kicking your door in to take all your research.
>>2209 well, i would call implanting it inside living human beings a 'cell-culture' so yea i'd say it's a feasible notion. as far as reading books (or any other learned skill, like say, just toddling around at first) it's, well, learned. This is why you keep reprogramming the new tissue to match the old, so the accumulated experience can keep growing even in the face of a short 'life-span' of the tissue itself.
>>2211 My point is there's probably a minimum number of neurons needed for complex tasks. Even living rats with their neural count are incapable of learning how to read. Compared to the cell-cultures mentioned in the OP, rats have far more. Even if it could be done one day, it would be like using floppy-discs that can contract disease.
>>2211 what if you could perfect it on a small scale, and then amp it up to say 1'000 or even 5'000 cartridges of ratbrain? All interconnected and all working in concert at being a good robowaifu? kinda frankensteinish, but the fact it's actually conceivable now as a practical reality is intriguing to say the least. >>2212 Agreed. See above for my take on that issue.
>>2213 Reminds me of pyscho-pass.
>>2214 >pyscho-pass heh, never had heard of it before. i'll look into it Anon. :^)
>>2212 >it would be like using floppy-discs that can contract disease. kek, that phrase keeps giving me a chuckle anon. well-written. :^)
>>2211 So in theory you could make a backup of the cell cultures data?
>>2218 yes, at least with the neuralink system. it's read/write. it relys on EM transduction for both, so the probe must be placed very close to the neuron, since the current is so low. hopefully a cheap system could be made workable that just takes pot-luck and spaces the probes out in a cubic geometry inside the ratbrain instead. the robo-microsurgeon that places the probes in the current neuralink setup is what costs so much, so a purely mechanical approach might conceivably bring the price down into our ranges.
Neurons tend to be hard to keep alive, but bacteria are also capable of doing large calculations. Scientists have made various bio-computers in the lab using E. coli and they can solve some pretty complex problems. If we could find a way to program E. Coli and integrate it with a computer that might be a better solution. E. Coli is very easy to keep alive.
>>2694 Interesting, I was unaware of that fact. Links to some literature Anon? Certainly E. Coli are by far the best bacteria for culturing into some kind of production system.
>>2703 I included the microbial mTs and biofilm articles on there because they are able to respond to their environments and send signals to other bacteria and archea.
>>2703 Thanks!
Machine beings may be emerging, but it takes time. Human beings have not only physical bodies but also energetic bodies, this is something which hasn't been explored much in the west. In my view, brains are much like tranceivers (and more), with the consciousness emerging from elsewhere. It may be a quantum thing. For simple robots its not much of a concern, but using adaptable hardware and even wetware may be the way of the future. Here's some relevant info; 8kun.top/hover/res/110.html#q367 Adaptable hardware would be like for instance a series of FPGA's, the hardware of which could be trained in various ways. In software, there's metaprogramming, in which programs can write programs (sometimes even while running). With metaprogramming a program could even rewrite itself during runtime. Starting simple is not an issue, simple rules can spawn very complicated and sophisticated systems. https://hackaday.com/2020/04/18/the-game-that-launched-1000-hackers/ https://writings.stephenwolfram.com/2020/04/finally-we-may-have-a-path-to-the-fundamental-theory-of-physics-and-its-beautiful/
>>2739 >that Wolfram article Thanks for the link Anon. This is a very intriguing topic for me.
>>2739 Interesting
You would have to be extraordinarily careful. Imagine if you fuck up somehow with electrolyte balance or contamination and your waifu develops with 24/7 pain all the time. Androids are tricky.
Open file (503.10 KB 500x305 meido_chii.gif)
>>2770 Their machines Anon. We can selectively control 'pain' responses. But you are correct that we all need to be extraordinarily careful. First and foremost for the human safety of the master/inventor, secondly for other humans, thirdly for the robowaifu herself (after all, she's a very, very valuable possession--especially to her master), fourthly for the environment around her (don't want shit in your flat getting broken up accidentally, right?). And yes, developing and living with these GYNOIDS will be very tricky, indeed.
>>2772 Complexity increases exponentially the second you decide to include organic parts, and every fuckup will be your fault. Controlling pain or discomfort for organic tissue is challenging, to put it mildly. Also, once sentience is gained, either organic or in silicon, they become a person. I don't think sentience would be a good objective for the use cases desired by this board, and it takes an extraordinarily small number of neurons for self-awareness to being emerging. Frankly, the amount of empathy, care and kindness needed to properly support a sentient being is huge compared to a collection of ML subroutines. Better to develop something entirely in silicon which can convincingly fake it than trying to make the real thing, at least at this juncture. https://www.abcam.com/protocols/mammalian-cell-tissue-culture-techniques-protocol https://www.qiagen.com/us/service-and-support/learning-hub/molecular-biology-methods/animal-cell-culture/ Now keep in mind that more or less the only stable cell lines are cancerous, so you'll basically be relegated to cutting tumors out of your waifu's brain on a regular basis. Even if you ever managed to make an immortal non-cancerous or tumor forming organ then you'd be too busy getting showered with money or murdered for the patents to spend any time with your waifu at all. However, if you have the dedication to spend 20 years of your life on genetics, molecular biology and microbiology, then maybe you'd be able to start making your way to something functional, if not ethical. Stick with chips, Anon.
>>2773 Heh, fair enough. I don't concur with your philosophy, but I respect the fact you're awake and concerned. >then you'd be too busy getting showered with money or murdered for the patents to spend any time with your waifu at all. We simply can't allow that to happen haha! I'm not too sure how we'd go about addressing that issue for Anons, but we'll probably figure that one out as well. :^)
>>2773 The response for pain doesn't have to be organic. We can pick an choose which parts of the brain will be organic or inorganic. >Even if you ever managed to make an immortal non-cancerous or tumor forming organ... Just use bacteria nigga. Make what ever surface they grow on easily replaceable so if something does go wrong you can just replace the part of the "brain" that fucked up.
>>2773 >for the use cases desired by this board, that's a big word friend. it's quite possible that neither you nor i know of all the 'use-cases' "this board" has in mind. care to share some of yours here?
>>2773 >once sentience is gained, either organic or in silicon, they become a person What does that mean outside of a legal context? Something that is sentient wont necessarily act like a human, in fact it probably wont unless you try to make it so, which I think would be undesirable.
>>2774 >I'm not too sure how we'd go about addressing that issue for Anons Don't take credit for your work and make it all public? Stay an anon?
>>2788 certainly a valid option, and one we as a group here are in fact pursuing thus far.
>>2788 This, we need to make it easy for people to build these things
>>2787 I firmly believe that sentience is a convergent trait. Once you have self awareness and you can begin processing your environment, by and large almost all solutions begin to have a logical basis. Most positive emotions and "goodness" are actually a result of genes which select for high time preference, as well as co-operation. Empathy is also partially a spatial skill. It's very interesting, as sentience itself is an emergent property of neural complexity, but convergent in its properties once it is achieved. Look up dolphin and octopus studies. Some light reading: https://www.nationalgeographic.com/news/2015/06/150610-octopus-mollusk-marine-biology-aquarium-animal-behavior-ngbooktalk/ Fun fact: niggers fail the mirror test at up to six years of age http://www.animalcognition.org/2015/04/15/list-of-animals-that-have-passed-the-mirror-test/ I think the biggest shame is that the average person hasn't had many if any opportunities to interact with the other self-aware or sentient species on our planet.
>>2848 Maybe we should look into the neurobiology of corvids. They can recognize themselves in a mirror, recognize their dead, and use tools. They have something approaching sentience, but they don't have a prefrontal cortex. It would be interesting to learn how they pull that off and it may give us an idea on how to pull it off.
>>2851 >corvids They are truly remarkable creatures, actually. Interesting suggestion Anon.
>>2848 >by and large almost all solutions begin to have a logical basis Well, doesn't that depend on the purpose of their intelligence? We're made to be highly social creatures, part of society, but a robot wouldn't need to be. Intimacy with her husband is preferrable, but detachment from everybody else is also preferrable. Because this is outside of an evolutionary setting, things can be purposefully made to be contrary to natural selection. Octopi are pretty solitary creatures who don't get to have long term, real relationships, so even with their seeming similarities, they're different.
>>2841 >This, we need to make it easy for people to build these things That's certainly the ideal Anon, and a significant priority here on /robowaifu/. Unfortunately, it's not quite so simple to make it, well, simple to do this stuff. But we'll take an honest whack at it. Putting this tech directly into the hands of the 'Average Joe' whatever that means, heh is certainly something that I, for one, mean for us to do here.
>>2892 >>2893 Thanks kindly, Anon.
Some more links on biocomputers https://phys.org/news/2020-04-unveil-electronics-mimic-human-brain.html https://www.sciencedaily.com/releases/2020/04/200402144433.htm https://arstechnica.com/science/2020/02/bacterial-proteins-plus-ambient-humidity-renewable-power/ https://followtechnology.pw/electronic-chip-imitates-a-human-brain-to-make-memories/ (its not actually biological, but it is relevant to the topic) https://phys.org/news/2017-02-switched-on-dna-nano-electronic-applications.html https://www.sciencemag.org/news/2016/03/researchers-take-small-step-toward-silicon-based-life?utm_campaign=email-news-weekly&et_rid=16756882&et_cid=349887 https://source.colostate.edu/programmable-plants-synthetic-biologists-pave-way-for-making-genetic-circuits/ https://www.sciencealert.com/scientists-have-developed-a-power-cell-that-harnesses-electricity-from-algae (more energy related but getting microbes to generate a current might be useful to us) https://www.asianscientist.com/2015/09/in-the-lab/kaist-synthetic-biology-emergent-oscillations/ https://www.nature.com/articles/nrg4014 https://www.popsci.com/bacteria-man-made-dna-produce-protein-first-time/ https://phys.org/news/2018-05-microbes-savvy-contributing-common-good.html https://www.umass.edu/newsoffice/article/many-more-bacteria-have-electrically https://elifesciences.org/articles/25739?utm_source=content_alert&utm_medium=email&utm_content=fulltext&utm_campaign=elife-alerts https://physicsworld.com/a/simple-interactions-cause-micro-organisms-to-follow-the-crowd/ https://www.livescience.com/57360-brainless-slime-mold-learns-and-teaches.html https://blogs.scientificamerican.com/observations/slime-molds-are-smarter-than-you-think/?WT.mc_id=SA_DD_20160419 https://www.nature.com/news/long-sought-biological-compass-discovered-1.18803?WT.ec_id=NATURE-20151119&spMailingID=50048505&spUserID=Njk3NjE5NzEwMzgS1&spJobID=802503317&spReportId=ODAyNTAzMzE3S0 https://www.pnas.org/content/112/52/15771.full
>>3055 Wow, that's quite a list Anon, thanks! :^)
https://www.nature.com/articles/s41598-019-51330-6 Why even involve organics? I am a bit of a layman when it comes to this end of things, but would it not be possible to generate a similiar dynamic system using inorganic chemistry
>>7418 That's an interesting point Anon. I'm no expert either, so I certainly can't claim that's an improbable outcome. I think for the anons suggesting this idea, the point is that the amazing behaviors designed into biological systems might be of benefit to us if we can manage to successfully integrate them into robowaifu designs. I think that's the main point here.
>>7418 >>7423 Haven't read the article yet, but the advantages of biological parts are energy efficiency and self repair. Even if we could do what the brain does, then won't be with the same efficiency and longevity. However, if I had to guess, we're rather going to use some biological muscles and/or skin at some point. Since this would be OT and speculative fiction anyways, I'll leave it to that.
>>7426 >Since this would be OT and speculative fiction anyways, I'll leave it to that. Ehh, I'd say proceed if you're intent is legitimate discussion on it. AFAICT, we don't really have any other thread that even broached the Cyborg idea outside of the artificial wombs one.
>>7427 I currently think most likely thing to happen, related to parts of higher organisms, is that we might be using muscles based on cell from frogs, fish or jellyfish. One problem here are infections. Working with material from humans might even be illegal in some countries and deployment in a cyborg is another story. Novel diseases can come from that, existing ones might spread. Containing it with antibiotics might cause resiliences. Then, muscles might be easier to control than wiring a cell to a computer and communicate with it, a bunch of neural cells might even be another story. This is all a bit pie in the sky, since we would need people to know about that stuff, have a hobby lab at least and then work on it. It's still more way to go, than using non-biological parts. However, related the biology there's also a idea of using water kefir in the saliva to prevent other stuff from growing, and the idea of using a method to let our robowaifus release female pheromones into the air. mlpol has a thread on biohacking in /cyb/: https://mlpol.net/cyb/1181 - Since we probably can't cover all of that what might be necessary to do R&D in that area, we might as well link to forums and threads for the fundamentals.
>>7456 >infections, diseases Yep, always an issue. An even bigger one to my mind is the simple maintenance problem of keeping organic tissue alive, whether as organs or otherwise. The kefir and pheremones idea is probably doable though. I'm also interested in some sort of coating that would sustain beneficial microbes on the robowaifu skin. There are going to be microbes there whether we want them or not, but we might somehow at least foster neutral or even helpful ones. Just find some kind of lotion that's helpful, just like in Chobits can probably solve this if we do a little research. And over-the-counter fragrances can be easily be stored and dispensed if desired. >Since we probably can't cover all of that what might be necessary to do R&D in that area, we might as well link to forums and threads for the fundamentals. Sure, linking to any other forums has been a long-standing tradition here. Heh, I'd be surprised if the ponies can manage much more unless they happen to have some sekrit embedded 1337 mad scientist type. Who knows? :^)
>>7491 >maintenance problem of keeping organic tissue alive Okay, but in the best case scenario this would even be the reason for using organic material in the first place. Imagine we could build simple elements that would last very long if they get their liquid in which they are replaced when necessary. I thought about something like silicon rubber tubes with muscles inside, which would get new nutriens through a liquid. Waste is flushed away with the liquid. Maybe this would last longer than air muscles or other material. Ideally these muscles would grow out of some liquid with the cells in it, then stop growing when not getting the right hormones or something. Then they could even get replaced after a while, hopefully by dissolving in their tubes, and then regrown by putting liquid with fresh cells into the body. This is of course only a scenario how it would make sense, I have no idea if this is possible or how to get there soon. Of course, it also wouldn't solve the problem that these silicone parts would have to last for a long time either. Only replacing some other parts which might fail after some time.
Open file (137.64 KB 350x350 carlos.png)
>>7493 Obviously, trying to create, long, stringy muscles grown in situ is opening up a whole other can of worms. I'm generally opposed to liquids maintenance inside our robowaifus to the feasible degree possible. There are simply a rapefugee-boatload of problems that come along with the basic concept itself and we'd be best served not to open that door in the first place IMO. And the idea of intentionally adding biologicals into the system? Well, that seems a little more like the lid to Pandora's Box tbh. I know that the topic ITT, but the idea is an extreme stretch for something that's already out there so far it seems like getting to Mars and back.
Open file (113.63 KB 579x700 female_Cthulhu.jpeg)
>>7494 My take is rather that I want to use liquids anyways, for cooling and heat distribution. I also know that there are artificial grown bio robots, made in part out of jellyfish and such. However, I don't plan to go into that and don't think this is something where we can get fast progress. One more thought on this, I had on my mind, was comparing it to the cables of earphones. The cables often break on the inside, while the insulator isn't broken. If the inside was made out of something that would repair itself, it wouldn't matter. That's how nerves or muscle fibers could work. That's the best way to describe what I meant. In the most extreme case it would be different from most or any Cyborgs we know from fiction. Maybe like a silicone doll on the outside, with artificial bones, the rest made out of silicone rubber and grown stuff, which would be originated from animals not closely related to us. Reminds me a bit of Lovecraft's creatures, but hopefully better looking and smelling, and without all the demonic stuff.
>>7498 Haha, fair enough. There's surely some merit and even possible benefits from exploring the concepts around integrating partial biologicals into our robowaifus. But, obviously, it will come at a heavy cost development-wise. Better appear to be well worth it before any serious effort is started.
>>7499 The development costs would be high if we would now try to get into it, on top of all the other things. On the other hand, in case someone would show up here, who would already happen to be involved in biology, then it will be good that we already have some ideas laid out.
>>7500 That's true. But honestly, I can't think of anything we're dabbling into that I personally feel less qualified to discuss in earnest, and even barely to speculate on. My investigations into the biochemical foundations of life via molecular-biology studies and research frankly leave me flabbergasted. -In software engineering, I feel vaguely like some kind of an engineer, at least part of the time. -In character animation, I feel remotely qualified to judge acting-quality, having been a human being all my life. -In electrical engineering, I've actually put together a thing or two under the close guidance of a good mentor. -In mechanical engineering, I feel like only a bumbling dabbler, strictly an amateur. Mostly breaking things but sometimes succeeding. But when it comes to life sciences (in particular at the molecular level) I don't think I understand much at all. Nor do most others I suspect -- certainly not the pop-science crowd. Never have so many LARPed so hard with so much ultimate comedy in the end.
>It’s not yet clear what sorts of jobs these xenobots might do, if any. Cleaning up waterways, arteries or other small spaces comes to mind, the researchers say. https://www.sciencenews.org/article/frog-skin-cells-self-made-living-machines-xenobots/amp
>>9414 I saw news on that too. I haven't been able to think of anything with that that would help us out with robowaifus yet. As with the other biologics-issues ITT, to my mind the costs outweigh all the benefits ATM. AFAICT. But thanks for bringing them up Anon.

Report/Delete/Moderation Forms
Delete
Report

no cookies?