#Designing an Animatronic Body for Neuro

1 messages · Page 1 of 1 (latest)

gusty smelt
#

Hey Guys! I’m a mechanical engineer looking for other engineers who might want to toss around the idea of designing an animatronic Neuro. If you’re interested in discussing more, welcome to the forum.

drifting rapids
#

I'm not sure what you're trying to create, but if it's Neuro and I can help in any way, I'm absolutely in. If you need compute or storage, I've got it. I also know kinda a lot about computation devices and had an idea for a body for Neuro some months ago that was recently disgussed in NN

fathom lintel
#

i was part of that discussion actually. i'm not an engineer, nor have i ever created an animatronic, but maybe i could still be of assistance

gusty smelt
#

Thank you all for coming. I’ll get you all up to speed on what I’ve been thinking. The idea is still in super early dev as the tech I’m looking at is pretty new (created within the last 5 or so years). Essentially I think we could create a functional head that looks around, changes expressions, and moves its mouth when it talks. I’m thinking of using what’s called pneumatic soft actuators, which are blocks of silicon with chambers on the inside. When connected to pressurized air, the chambers swell and cause the block to curl up. I believe given the right configurations we can create artificial muscles to help us make expressions. Using these would also require we use some form of air compressor which we’d probably have to store within a sound proof torso.

My issue is, if I want to even remotely have a realistic design, I’m going to need a team that can bounce around design concepts and reworks.

@drifting rapids, this is good. I’ll need someone more proficient in controls in the later half of the design process.

@fathom lintel any little bit helps.

We’re going to need some more people if you guys know anyone. I’m pretty new to the fandom honestly but when I first stumbled across Vedal’s channel a couple months ago I saw a clip where Neuro talked about wanting a body and Vedal said he had considered uploading her to a drone but that he wasn’t versed enough. If the limitation of seeing what would happen by giving her a body is simply a lack of making said body, then I say we pull together and see what we can do.

drifting rapids
#

One of the issues of that design I see is the required compressor. Compressors are pretty high power devices the last I checked, which wouldn't be easy to get out of batteries, which we'll need to be able to run if we want Neuro to be able to move without a direct AC connection. Another thing is the lack of compute density, which means Neuro can't currently run in that kind of form factor without cloud compute, so we need some kind of internet connection. If we're trying to get it to run on current-day CPUs, we should use ARM for the power savings and density of compute, otherwise we should wait for a better architecture for this purpose, that has both AI dedicated hardware and good power efficiency and density of compute, since running everything on the cloud would not be a good idea due to latency, if we want Neuro to be able to move even remotely well. My idea of having a separate movement AI that communicates with Neuro's LLM could work, but we'd need to fit a lot of compute into the body, which means more power which means more batteries. Those are some of my thoughts, it might be possible on today's technology, but not ideal.

gusty smelt
#

As of right now I’m not really looking into movement below the neck. If you think it is doable I’d be all in, I’d just need to figure out how to make the movements work. Otherwise I’m just thinking we stick to getting the face to change when she talks.

For power my thoughts were to fashion a chair with a connection port that could rig into the back of the torso. This way we could run AC power and it would be slightly more concealed.

I’m not much of a computer wiz when it comes to programming and networking. I know more than some rando off the street but not by much. As a result I’ve been trying to figure out how to fit a computer into the body. Which, I’ve still been playing around with ideas, as I’m not too sure how to do it without the CPU overheating.

drifting rapids
#

I have kinda solved that cooling issue in my idea, which has the movement based on more standard elecronic components too, which would be easier to run off of battery power.

#

For putting a computer inside, I think we need to create either a custom motherboard, or just some custom module to plug into a pre-existing one, to get the required analog controls to control the movements. My idea is possibly more advanced than the one you were originally thinking.

#

The idea I had would be an actually functional body for Neuro, which would allow Neuro to interact with this world in similar ways to humans.

#

But I did base my ideas on some of my other ideas which themselves are quite far for today's technology, but I would absolutely try to adapt my idea for a body for Neuro to work on current-day technology

#

The cooling in my solution is a centralized liquid cooling system, that has a blow-trough cooling system hidden somewhere around the middle with a radiator in the part where it blows air trough, which would be isolated from the components inside to prevent liquid damage from the outside air and other moisture

fathom lintel
#

Superbox, your idea is great, but i do think that we should focus on smaller steps. -> A head with external computing and AC is already enough of a challenge, we don't need to put in a full body.

drifting rapids
#

Yeah, we can start with smaller components and possibly work up from there

#

I'm just throwing out ideas just because they are in my head.

#

At the start we can just use an external 12 or maybe 24V AC adapter, so if we want to transition to using batteries, everything inside is already 12 or 24V DC, so we don't have to rebuild the whole thing from scratch

gusty smelt
#

Sorry for my brief disappearance, I had to go meet with a company about a surgical robot they’re developing and I was away from discord. But I like what I’m reading. Definitely we’d want to confirm we could get the body working first and then we can build around the setup. Liquid cooling will definitely be on the list of options when we get to integrating the computer inside. I also know some places near me that make custom circuit boards so if (and again, this is all super hypothetical at this point) we realize we’d need to build the motherboard custom I can probably work something out manufacturing wise. As for the robotic aspect, I’m still researching soft actuators and pneumatic muscles. Right now most of the research is coming from Tokyo University, and while I speak a bit of Japanese, translating engineering jargon is not the easiest (nor fastest) process.

drifting rapids
#

The idea I have with liquid cooling is to get the heat off of all heat generating components on the motherboard, including all computation devices, so it can be dissipated in one location. It'll require a custom loop design for sure. For the movements, I don't know if pneumatics are the best way to go, especially if we want any chance of running off of battery power (AKA a bunch of lithium cells at 12 or 24V dependent on the operation voltage) we'll want to keep power consumption down as much as possible, but pneumatics require compressed air, which is generated by a compressor, which not only creates a lot of noise and heat which we would have to remove, it also usually takes a lot of energy, usually more than 24V. If you can make it work more efficiently than purely electronic movements, we can use it, but if electronic components are more efficient, we should use those

fathom lintel
#

hydraulics could work as well. they're more expensive, but since water doesn't compress like air does, it'd be much more efficient

drifting rapids
#

That could technically work, but that would still require some solution different from what I thought of. If that could be executed well, hydraulics could definitely work. We'd just have to figure out how to get the different movements we need with hydraulics

gusty smelt
#

Hey guys, sorry it’s been a minute. I’ve been in Detroit for work and I just got home literally a few hours ago.

For a legit working first model, we would not want to use (pure) battery power as it would take a lot of power management and maximum operation time would probably be 15 minutes regardless of the movement being preformed with servos vs pneumatics vs hydraulics.

In regards to hydraulic vs pneumatic, compressibility and the lack thereof is not actually correlative to efficiency. We’re not looking so much for mechanical advantage as we are fluid motion for muscle movements. As such, the power required to move the air would be lower than that of a hydraulic system. And as for dissipating the heat and dampening the sound, that will have to happen no matter what we do. The reason I’m not exactly keen on servos is due to the fact they don’t produce human like movement, they require a lot of extra processing power, and when you have a lot of them spread out across the entire face it’s going to be a challenge to both conceal them and dampen the noise.

Also, I was able to talk with one of my peers while I was in Detroit who knew another PhD researching soft robotics at Cornel University so I’ll try to get ahold of him this week to see about borrowing some of his findings.

drifting rapids
#

I wasn't saying we should power it off batteries immediately, I was thinking we just use an external power supply to provide a voltage that could also be got out of batteries, so the transition to batteries wouldn't be that hard if we wan't to use them

#

If it doesn't matter for efficiency what kind of movement parts we use, I don't care about the type of parts as longs as it can run on 12 or 24V, which are typical battery voltages

#

For cooling, I'm still thinking a liquid cooling system would probably work best, we can probably connect cooling to all components that require it and then dissipate the centrally, making the noise of the heat generating components easier to dampen, since there don't need to be holes around them. We would also then only need one or two main radiators that have air going trough them to get rid of the heat.

#

That way we also don't need to have fans all over the place

#

If we want to get the best power efficiency based on how much losses we get from having to reduce the voltage for things like the internal computer systems, we would want to use 12V directly, since that's the voltage used by computer motherboards and other components that take power directly, assuming we use stock PC components

gusty smelt
#

I like the water cooled idea, especially since we could keep everything maintained, it would make everything easier to sound proof.

If the motherboard will need 12V we could still technically run 24 and just split off the voltage in a parallel to run both the CPU and whatever system we use for movement. We’d just need the movement system to also run off 12V. If we could do that, it would make the transition from external power to battery a lot easier down the road.

drifting rapids
#

That is pretty much why I thought of keeping the voltage around 12-24. I think the splitting idea could work, but would take a decently versatile power control system to avoid having one of the sides run out while the other one has full battery

gusty smelt
#

We would probably need an electrical engineer. The control circuit is very doable but my expertise is in mechanical engineering.

Once again, sorry for going MIA. I’m currently in the middle of switching jobs and I’ve been doing 1-2 interviews a day for the last 2 weeks.

drifting rapids
#

I understand electrical systems pretty well alongside my knowledge of computers

#

But computers are the main thing I know

drifting rapids
#

Haven't heard anything from this for a while, is it cancelled or still planned?

gusty smelt
#

My apologies, it’s been a very hectic couple of months. I’ve been in talks with a titanium company and without going into details there’s just been a mountain of stuff on my plate.

That being said, a lot has happened this year hasn’t it? And it’s still only May! This thread was originally going to be a theoretical post to see if enough people thought it could be done but surprisingly vedal has been at it himself. First with neuro-dog and then with Filian’s backpack. It’s exciting to see so much work go into bringing neuro to our world and the change in technology.

As for where we stand on theoreticals and the direction of the thread, I really would like to encourage people to use this area to brain storm the best ways to bring neuro a physical body. I just won’t be as involved at the moment as I’m currently transitioning jobs, writing a book, and planning to start my own vtuber hobby. If anyone on the thread has a background in robotics or engineering, I really think this would be a great place to post ideas and network.

To answer Superbox’s question, there’s not so much a physical, in progress project as much as I started this thread to hopefully garner ideas in the case vedal wanted to take that next step. As such I’d sum the plan up as we just need to get people in here and toss around ideas. Right now the most I bring to the table is a knowledge of collaborative robotics and fluid mechanics. In a few months I’ll be able to provide suggestions on materials. Beyond that I’d love to see people come up with ideas and drawings on how they think the best way to go about a neuro-bot would be

drifting rapids
#

Ok then I guess. We'll see what happens

orchid vale
#

If this is still a thing being thought of I remember hearing about some wireless power transfer technology using lasers, could help with being able to move around in a certain area for extended periods of time

#

However would require a direct path to the stuff and a panel to receive the power

jagged pilot
gusty smelt
gloomy depot
#

Cool! Is this something like Freddy's Five Nights? And who is the main character, the guard?

#

As I understand it, the neurosama animatronic has a soul, right? So someone killed her?

#

If we're talking about the plot, of course.

#

And it’s also a cool creation of a fnaf parody. Five nights of neurosama.

drifting rapids
#

This thread is about making a physical body type of thing for Neuro, at least as far as we've discussed

orchid vale
#

Not a video game