#Neurobot-1000 3d model project
1 messages · Page 5 of 1
nice idea. I like it. For now it could work
i just realized, that vid showing tentacle movement could be used for a tongue; Wire pulleys maneuvering flat plates to curl in three dimensions
i'm fairly sure that's how the human tongue works; Four muscles connected together
when one contracts, it pulls the other three diagonally towards it
Could be cool. a tongue would allow her to do funny expressions sometimes
yeah fr
sticking her tounge out when vedal turns his back after saying something
If tongue then where speaker?
The tongue is used for a lot of mouth sounds, but if there was actually a speaker producing a voice from the back of the mouth, the tongue would interfere with that. It could make some noises harder to make or easier though.
maybe stick it just under the tounge?
A small speaker in the mouth may be fairly weak too, but a larger one in a cavity in the body with a tube going to the mouth could be much larger and more audible
just use speaker imo
i don't trust today's material science to produce viable sounds and imo it's beyond me really
give her actual vocal chords fr
a voicebox and proper cheek deformation is required and even state of the art still sounds haunting
less material science and more if the AI can handle it. I'm pretty sure small transformers could, but it's a lot more work
HMMMMMM
if we can find the right material and i somehow replicate EVERY muscle, then it could work, but that's a stretch
Honestly, maybe we should create multiple neuro versions?
yes
do not reccomend
first neuro version should be the simplest most MVP version
That way we can keep feature creep out of the initial stuff
potentially neuro will be around longer than you or I will be alive, so we have a long time
No teflon membranes or massive speakers in the first version
yeah so long as it runs im satisfied
i'm thinking that a lot of what we're discussing rn, we'll look back on with hindsight as being suboptimal, but that makes me excited for the future
idk what we're doing wrong rn but in future versions we'll be able to correct faults and iterate
very exciting
We will never know what's wrong till we build her somehow. And even then, we will miss something till she runs. And even if something doesn't fail, there's always room for improvement
Gotta make sure our ability to make stuff increases over time, so that should mean making tools or getting good bases or libraries that can be used for general stuff
So, if we use an engine, we should record how well it works, battery configurations, etc.
Maybe youtube idk
make it easier at least
ye
or in the roof of the mouth if it doesnt mess with the eye mechanisms
she needs at least some level of speaker
then the tounge would get away if we are going to add one
oh how big/loud is the speaker? Tongue usually attaches in the front of the throat so i was thinking we could use the back of the throat, nestled into the skull
ah idk
im just parroting sim here
Probably doesn't need to be that big, the main problem with sound quality is... (Well... it's another world that we could talk for ages) well... size of speaker, shape of the back of the speaker and direction of it.
If it doesn't sound like 2 dollar headphones.. we're probably good
...maybe a speaker air-sac connected somewhere in/around the mouth area by a tube isn't a bad idea...
if we got neuro her own vocal chords she could have her own voice for real
V4 voice lets gooo
nice idea
Heck! That skeleton style looks crazy good!
O, that's kinda cool: https://www.sideshow.com/collectibles/mark-i-endoskeleton-molecule8-903050
I think it's a male skeleton though
it likely is
Jesust! It was on sale?!
we wouldn't want it
that exo was cheap as fuck
To be fair, it was an action figure to scale. so... it's just a collectable
I've been abalyzing all commercial robots and i'm noticing that the eyes are the largest hurdle that people have with uncanny faces, followed by skin slackness and bad proportions
so my priority must be on the eyelid mechanisms and alignment
true
a big part of it is also complexity in the eye movements
lack of tracking leads to a dead expression
too much with a dead face, leads to a predatory expression
Yeah, facial expressions are generally hard to get right for humanoid robots.
That animatronic head made by Matt Denton posted earlier did pretty well at it with the eyes though.
yeah. Seems like beleivable expression goes hand in hand with eyelid tech
another problem with the robots is that the skin material always bunches up unnaturally when it's making expressions. We can avoid this by abstracting it away
I’ve seen people struggle more with the mouth honestly. 99% of the time the mouth movements look uncanny.
In theory
I think the NaVi animatronic comes closest to achieving natural mouth movements.
i like their choice to use individual motors to stretch the skin where the muscle would bulge or pull on it, it makes it very reactive and lifelike
i don't like their proportioning or their treatment of the angle of the mouth though
Yeah
As we can see, six muscles on both sides of the face, so 12 in total, pull on the angle of the mouth, and this causes it to be 'extended' and bulged outwards under the skin in a healthy face
That's... that's a lot of movement. No wonder why it's so easy to mess up an artificial face
This is a very exagerated example, but see how the muscle mass and tension pulls it out sideways? Each of those muscles is connected to a certain type of expression, so it flexes a lot on a healthy individual: Sideways for satisfaction, upwards for a grimace, diagonal for happy, downwards for sad, etc
it's even more pronounced on males
Yeah not a lot of people get it right, the newest Disney Audio-Animatronics are the closest I’ve seen. There’s also Ameca but that’s uncanny as well.
i don't understand why nobody emphasizes it. It's a robot so we can afford to overemphasize health indicators and make people associate the robot with being healthy
they need to hire sculptors
people who can create a face in any expression at any angle artificially
anyway my plan is to manipulate these areas of the face the way the body does it, using motors pulling cables
i'll try with an abstracted version first without skin, to see if i can get that to look passable
i probably won't go for skin if i don't need to, as it's another source of dissnance
people can tell when the skin doesn't move right
these are all the key expression muscles
sounds pretty good. I hope you find the space for all of the motors required. It sounds like pain to plan out all 
Yeah, I want to learn how to sculpt in silicone but it’ll probably take me a good number of years to become proficient enough. I mostly just sculpt in clay.
fr i might need to embed them into the skull itself
I feel like the face needs to have skin
idk how you can have a face without skin
i think if we do it right, we can use metal plates that have the dimensions of skin, this will make it look like a moving sculpture
Heck! That's an amazing skill already!
I saw something like that. It looked freaky tbh
i wouldn't want gaps like it had
No, not the avatar thing. I saw something way more detailed online... if I could find it...
this kinda look
Ah, that has skin though
That seems like it would be even more uncanny
how horrifying
Whoot? "Living skin"? Is this organic skin?
I was going to post this as a joke. Turning her into a cyborg would be kind of weird. 
Yeah they grew skin cells and put it on a robot
Literal terminator-esque cyborg.
It’s kind of gross though tbh. Idk.
i’m not an engineer and i’m enjoying the project but i’m bothered by the age and height evaluations of Neuro
Yeah... is like... on one hand... my mind is like: SKIN FOR NEURO! YEESS! But... on the other hand... needs way more polish, probably. 
There’s the risk of the skin rotting and that would be really really nasty. 
the head height is based on measurement against real life human heads in the second row as well
Yeah, she would need to constantly bath her face in amino acids so that it doesn’t rot off.
Gotta feed the skin with food, oh god! And something would need to regenerate dead cells
sounds like a nightmare
it needs to look natural in a human environment
Yeah I honestly agree
Depends on the art style
i've found that the head's a bit too wide as well, so i'll be narrowing it a bit
Some anime have adult characters with big heads
Yeah like I said, it would be gross.

I always thought neuro's most "accurate" height was 145cm, but, this is an anime character and every vtuber says, compares and zooms their character however they like, and we don't even have a "cannon" height in place so...
if 121cm matches her head to body proportions... good for me. It maches more her personality too.
She's childish and looking like a 12-13 year old is danger territory of teenager and a puberty body she doesn't have. 
Finally found the robot head: https://sketchfab.com/3d-models/head-of-robot-tima-0a21c1001aeb4499a04099447fc16823
Here is the mesh for test of riggid FACS system and NPR shader I did months ago.
I found Riggid FACS system will cost so mush that will lag the real-time rendering/gaming and run out of my time to model, so I won’t detial face like this in my new project, but I still love the though and will reach that level once I have enough budget or faster w...
ew im not gonna do that rest assured
i don’t necessarily think it’s safe to make her too young however, both because of Neuro herself and other reasons
Honestly doesn't look that bad compared to a lot of other robot heads
She does act around 12-13 most of the time tbh.
13-14 maybe
it has some aspects to it that i like, that being the flat metal plates covering exposed internals without needing creepy skin
With ADHD
Oh gosh... yeah... ADHD sama...
Also true. It's... a delicate balance...
It's also a 3D model you can download and has CC Attribution license
Part of why she’s relatable to me 
Also we need to consider that Neuro will probably mature by the time this is made. Vedal talked about her being serious enough to organize her own collabs. She probably won’t be an immature baby AI forever.
Very very good point
and that's a pretty hard dilema. Either we ignore this fact and we stick to her design proportions (which would make sense) or re-scale the body so it doesn't look too weird or even redesign her body to be more "mature" but I don't like the idea of entering that territory.
(People probably want to see neuro as how she looks now)
Yeah, I was thinking maybe decide on the body shape after her new V4 model comes out, so we could get a better idea of how she’ll change over time. Then again I think the people working on this might get bored, I feel like 4’8 is a safe bet, still shorter than most people but it won’t be too jarring if she matures.
V4 model
Evil hasn't got the V3 model, yet, that's too much copium
Yeah that’s assuming this project is still around.
Like I said, people might get bored.

Is... it's very posible this project stays. I always thought of neuro getting her humanoid body not a matter of "if", but as a matter of "when". This project just gives me the light and the confirmation that, neuro humanoid is... closer than we all thought.
Realistically, her first humanoid body will be from Ellie.
of course
True. And that's good because she needs to learn how to walk with two feet so, if ellie gets the green light and does it first, neuro has a chance to train on a what will probably be a 3D printed humanoid
Instead of failing to the ground hundreds of times with ours
(700 fuel refuels later...)
hahahahaha
Anyways, time for bed for me. Keep cooking. This is getting lots of progress 
i wouldnt say a lot
we are still in the 0.(0)%-0.(9)% territory
but its going steadily
which i have a hard time beliving
that we are so low?
i think we're at least 5% in
I'm getting adds for robots now thanks guys
You're welcome.
Any good ones?
idk i skipped it
should we post progress notes some time soon?
not sure how we should format it, since it's not like we have an update channel
Gotta find the sites with better ads
That might be a good idea. I think everyone would need to add something to a doc about what their progress was though. I know I mostly worked on a gui and math, but idk about everyone else's
google doc?
Google doc might be the best for formatting. Idk how else to do-- ah, yep
Neurobot-1000 Progress Log MatthewBadger 25/01/2025 - 29/01/2025: 25/01/2025: Concept art/design work for the face, measuring facial recognition points on an idealized face to simulate skin deformation and facial expression of the robot head. Research into bulk metallic glass for use in skelet...
ping me when you need edit perms
Ah, I usually comment. However I could use some perms
ah, nvm wait no im confused
seems like i can'y give specific people perms
Usually I just open up commenting to everyone instead of just viewing, then I can accept or reject comments
done
@opaque quartz why is progress only counter from 25th of this month
started recording progress at that point
old stuff would require trawling through chat and also, some was a jumble of private research that i don't know the specific times for
Ah, I didn't realize there were more pages... Guess we'll need a table of contents.
The auto-gen table of contents follow headers. So ctrl+alt+1 or ctrl+alt+2 to make headers will help it.
oh nice
i might add a basic summary of previous progress for newcomores
also lets ping a mod for a pin
(I might have over-engineered this one a little bit...)
Ok done on that document
is that trillions of rpm?
No, it's a really precise fraction estimating a gear ratio of 2/pi, (here I'm entering the inverse gear ratio, so pi/2)
nice
@ruby lagoon we should probably order the document in terms of dates
that way people click the date to find progress by everyone every day
feel free to smoosh your progress in with mine
Ah, so date should be top-level, not author
I don't think I can organize my stuff by date too much, so I guess it will all be today
Also I don't know if I'll update that doc every day, maybe every week though
Ok, how about now?
Ah, you have to press ctrl+alt+1, ctrl+alt+2, ctrl+alt+3, etc. for the table of contents to catch the dates/people, not just bold
ty for your formatting help
yw
looks good
doesn't have to be often, just whenever you feel like it. The resource is only there to keep people up to date on progress
you can go several weeks if you like
I'm just going to put all my contributions down for today
aye
might end up shifting it to bunching up all the changes for a week into one header
once we have a week's worth at least
since it seems to work well for the first header
sounds good
Wait... I have a github for my work: https://github.com/SimLeek/freecad.actuators/commits/master/
ty for the link
Yw
Lookup the engine cycle for it
What is that price?! This is the expensive ingredient to glow fuel! And that's just for 1 mere little!
(The mix required is 5-20% of this so, this bottle either gives us 20 or 5 litters of fuel)
technically we don't need to add it for the engine to run properly
a standing idle should use way less energy than the engine provides even with less performant fuel
just methanol + a tiny percentage of oil is all the engine requires
oh. Then maybe it could work?
Okey. We'll have to test if the engine doesn't break or something (Or it even provides the necessary energy).
Otherwise, we could switch to a gasoline engine. But we could consider that after the model gets more progress to see if we even have the space for it.
The smaller I found is still the NGH GT9-PRO engine. I tried to find smaller ones but almost all of them are vintage collections.
It's been said already but if we're using drone engines we probably need to make sure there's plenty of heat dispersal
Yeah... or find a way to water cool it. Especially an engine designed for planes where wind is... well... very much guarantied
We could circulate it around the body to let it cool off... Wait those are just blood vessels
If we use liquid coolant we'd also get a decent amount of body heat as a byproduct
Water cooling is a method of heat removal from components and industrial equipment. Evaporative cooling using water is often more efficient than air cooling. Water is inexpensive and non-toxic; however, it can contain impurities and cause corrosion.
Water cooling is commonly used for cooling automobile internal combustion engines and power stati...
an interesting read
the main thing i'm worried about with water cooling is the possibilities of leaks, since a full tubular distribution system would get yanked around a lot
🔥 accidentally on purpose recreating bodily systems🔥
we could have the water cooling be segmented but that seems complicated plus we'd need more pumps
also we have to consider where the pump would be placed and how it works
Buy Water Cooling Pumps - Watercooling UK supplies a range of Laing DDC, D5, Alphacool, EK and XSPC Pumps
yep. Most reliable pumps are PC pumps or... fish tank pumps... (Actually, fish tank pumps are really bad)
i have to wonder about the danger of a leak when dealing with so many watts with high voltage and current
If we can magically get the operating temperature to be around 98 degrees (farenheight) we're golden
Water and electricity create a deadly combination. Contact between them can cause severe injuries or fatalities. Water is an excellent conductor of electricity, making it extremely dangerous in environments […]
Electrified water is a great self defense method
i wonder if there are any red synthetic fluids lol
there's colorant for fluids and are compatible with pc cooling. So yes, we can have "blood"
And there's also non conductive "water". One of them is mineral oil. Some of them are used to fully submerge pc's in liquid
Neuro: become human is realllll
Heck ye
It's calling for us
😂
Of course we should consider the efficiency of the coolant rather than just "ooh we should give it blood!"
on a more serious note we don't want to use any coolants that contain dyes since that always complicates matters
i'm thinking we can run a network of tubes around the hottest components to circulate the heat
the longer the tube, the less likely it'll spring a leak because there are less joins
so unironically it'd be useful to design some sort of unbroken network of one tube which doesn't restrict ROM
ohhh
yeah we want to choose the best rated, least toxic, and least conductive, affordable option
uups. I didn't know that. The dyes i know are added after the coolant. Maybe some pc ones are better than this
(I never water cooled in my life)
Another thing to search I guess. 
XSPC EC6 is a high performance, eco-friendly coolant for PC water-cooling. It's based on a blend of refined vegetable extracts with non-toxic corrosion inhibitors and non-toxic dyes. It offers superb protection for copper, brass, steel, nickel, aluminium and has been tested with a range of acetal...
this one's the clear version without dyes
non-toxic, non-corrosive, non-conductive
bro this bot is on my case
😂
long link: essaying
She's watching your career with great interest
Also one of the longest links I've ever seen
same
@prime rose pin pls 🥺
Ah, I think those links sometimes contain tracking information. You can just remove everything after the B008PA1G00 part, including the ref, and it should still work
Oh great moderator of neirocord, hear my plea, please grant us the mighty pin of organization
Okey, so... regarding colored coolant. It can work. There's just one problem and that is water cooling itself.
First of all, maintenance is required. Just like dust builds up in your fan after some time, a water cooled system needs some maintenance after some years of running. Maybe like 10 or so.
The main thing to watch out on water cooling is metal mixing.
If any fitting or anything on our system uses different metals, there can cause corrosion. And that's something you can try and will work till a year later when something corroded.
If we make sure that our pumps/distributors use the same metals for their fittings, we shouldn't have much problems, even with dye.
(Gotta check that mayhems coolant, one second)
I mean there's the known physics you always want to consider before doing stuff, and then there's costs of fuel in different countries, noise, etc., that you didn't even know to consider. We should record the former in design and the latter in testing.
Oh right. Galvanic corrosion is definitely important with liquid cooling. It has to be either aluminum or copper, but not both. (Otherwise you're accidentally making a voltaic pile battery)
It is. Especially because it happens so slowly
Yea, I didn't even think of the battery part until now. I guess you could get some nasty electrical differences if you used a lot of metals
I just realised Neuro could become an alcoholic depending if it counts as fuel.
xDD
It would likely murder the engine but she probably could
Methane is alcohol based so... probably counts?
She doesn't have a liver so she'd just explode
In the future they'll be adverts of this is what alcohol does to your engine
fats and oils also burn pretty well, so if she had a filter like teflon she could eat fast food as well
Fast food and alcohol = robot food
Let's leave that for 2000 shall we
fair
We could also just use colored tubing. Regardless... it's not gonna bee seen and, with metal skin, it's not like neuro is gonna get cut anytime soon.
Lel. She could go to greegs, finally
But... But the humanlike facade!
The coolant dye can probably hold up with some additives
But i have no clue about that
to be fair, would be a funny excuse for her to go to the toilet (RIP for the toilet)
Why does robot blood have to be red? We can just ask neuro what colour she'd like 
Rainbow blood!
RBG GAMER BLOOD!!
Good point. In Detroit become human is blue. So... yeah, why not?
Vedal having to buy 500 flavors of coolant
ideally the coolent used would be one with minimal corrosion and metal mixing
liquid cooled seems like the way to go due to the small space meaning we can't just stick fans everywhere and still reach the internals
We can probably give neuro a badass central fan or something, like ironman
Heh, mix the speaker sac with the liquid cooling so Neuro can breathe out hot air
Maybe the engine exhaust too
huh that's actually a brilliant idea
Biomimicry ftw!
brb writing this down
That's genius 
trying to research the chemical content of the emissions
formaldehyde is pretty toxic
in terms of emissions it seems more attractive than a lot of other fuels
for handling, it seems quite dangerous though
maybe we should be looking at alcaho- i mean ethanol rc engines?
it's only 30% less powerful than gasoline, which isn't as low as i expected
Discussion Can Ethanol Be Used In Model Engines Engines
yes and the safety factor
both fuels produce quite safe byproducts but have differrent toxicities and power outputs
everywhere i look says that you can swap out the methanol fuel for ethanol in rc engines
i wonder if glow plug engines are different
seems like we'd need an engine built for ethanol or mcguyver our own glow plug setup
ty vanorsigma 
I'll search for it. It's harder because the pages don't directly tell you fuel usage. Most of them use "glow fuel" as name.
Converted a hobbyking AP15 Yellowjacket glow engine to run on bio ethanol and castor oil.
Bio-ethanol is sold here as fuel for flueless fireplaces for 1.65EUR/L. I suspect it is denatured by adding a small fraction of methanol, however the bottle simply says 95% pure bioethanol with no other substances listed.
The castor oil was 5EUR for a bott...
ngl sounds like farts
This video is about OS 60F GP running on Ethanol and OS A3 plug
seems like it. I think the standard is using glow fuel for the glow plugs, wheras niche speciality ethanol glow plugs have been developed to facilitate differrent fuel use
Honestly, I still like Stirling engines: https://youtu.be/YCjwlSX-tsY?t=1567
In this video, Mike machines a tiny Stirling engine on a miniature lathe - to our knowledge, this is the smallest beam Stirling engine that has been published to date.
Drawings for this project are now available on our Patreon page: https://www.patreon.com/ChronovaEngineering
FTC disclaimer: This video was sponsored by Brilliant.
About Chrono...
Also if you search mini stirling engine they're ~20$-30$
Probably not big enough to fuel much though, might power batteries a bit, idk
so long as the hp of the engine is greater than the hp used in a standing idle or walking gait, the fuel storage doesn't matter as we can pour in fuel on the move as needed until the batteries are charged.
so the question is if we can buy or machine one with the necessary hp
as far as i know, nitro engines are the best power to size/weigth ratio available but the fuel is extremely toxic to handle
RC Car from the video (engine included): https://bit.ly/46FJDPn
Nitro Engine by itself: https://bit.ly/48YGCvb
Find more engines: https://www.enginediy.com/?ref=d4a
Extra 10% off code: d4a
I earn a small commission if you purchase something from the link above
Patreon: https://www.patreon.com/d4a
This engine here which fits neatly into my hand ...
True, but I think it's fine if we consider rest=recharging batteries. That way there can be fuel still burning in the engines, and it just slowly powers the batteries from 10% or so back to 100%
Then the engines only need to be... maybe 2/3 as powerful
that's a good point
maybe we can machine a stirling engine designed to run on ethanol then
it would only need to provide a fraction of the power like you say
if she's on 24/7 then yeah
she could plug in when she's at home though
i think one of the benefits of the engine is the ability to recharge in random places in the outside world
she only needs access to some kind of high purity ethanol
and a place to rest up for a bit
ideal power output would be enough to let her walk places slowly, but if we can't do that, idling or sitting is fine too
YOOOOO
I just realized we can repurpose the waste heat from the water cooling system to power the stirling engine
so any inefficiencies in the gearbox and motors can be sent back into the system as future power
i think with this in mind, stirling engine is the way to go
Hmm... can work when no fuel, but while that will add some efficiency, it won't actually power the robot because that would be a perpetual motion machine
I wonder how much it would improve the total efficiency though
yeah it only saves on power through re-using waste
it all comes from the fuel
it wouldn't be 100% efficient either due to radiation of heat into the environment when not captured by the stirling engine
in fact ironically, doing it this way might lead to use needing to thermally insulate the tubes in order to capture as much waste heat as possible instead of losing it to the outside air
i was planning on leaving them exposed for better heat dumping
I think liquid cooling tubes are normally thermally insulated actually. That is they're made out of insulating materials.
nice
Ah, yep, PTFE is fairly thermally insulating
not only are stilring engines energy efficient, they're also space-efficient
I mean, if you want a tube shape... though I guess you could have multiple tubes and make it look more like a normal engine
true but it's very narrow
true
we can dump the heat of the lithium ion batteries into it as well
for the water cooling, i say we enclose the stirling engine's reservoir in a larger reservoire that the cooling system pumps to
Tbh, I was thinking of saving liquid cooling anything until we got some new manufacturing methods, like a ton of robot arms working together, or micro-robotics. However, if there are a small amount of components that make a ton of heat, then it could be worth it there.
true
Here I show you my way to get more power out of the Rhombic Stirling engine and improve its reliability and durability.
With various measures such as optimizing the running surfaces with the piston rings and the roller bearings of the rhombic drive, I am trying to make the Stirling engine capable of continuous operation.
In this way, we can us...
seems doable tbh
this is awesome
gonna write this down brb
Huh. Friction is more important than I thought
In this video, we look at the design and operation of an alpha-type Stirling engine. We show the advantages and disadvantages of such hot-air engines and focus in particular on the important function of the regenerator.
The Stirling engine essentially consists of two cylinders. One cylinder is constantly heated and cooled. In a Stirling engine ...
YES. Do not ignore friction in practical physics or mechanics. That's how heat, wear, and other problems occur. But that's also why lubricants exist.
In this video, we examine the idealized Stirling cycle. This cycle consists of two isothermal and two isochoric processes. The efficiency of the Stirling cycle represents the maximum possible efficiency achievable for a given minimum and maximum temperature. This efficiency is also referred to as the Carnot efficiency. The Carnot efficiency prov...
No, that's obvious.
I mean I didn't think a bit of extra friction would completely ruin the Stirling engine designed in the "Stirling engine development for cogeneration" video: "The relatively small friction differences [...] resulted in power differences of almost 300 watts" That implies Stirling engines may be better or worse mainly depending on how much we can lower friction.
Those engines sound cool, but they need a heat source to run. Are we looking to use ethanol as a main heat generator?
(Is impossible that the heat of the batteries is enough, right?)
yeah
the reason to use ethanol is that it's an external fuel source seperate from the batteries and can be carried like a water bottle
this saves space in the robot
that's true. We can't relly on batteries for everything. We're still on the same track, but. How we're going to make ethanol... burn?
We can't relly on batteries for everything
I mean, we can, actually
i was thinking we could have some sort of furnace/burner touching the piston of the engine, supplied by an oxygen pipe
we could use entirely batteries but in that case, neuro would be forced to recharge and woudn't be able to exist indefintiely outside of Vedal's lab
That would be realistic... but sad... she would still be linked to a goddamm cable...
For charging. There are fully electric cars.
True. But not enough charging stations 
I wonder how much energy is used for throttling vs gained through pressure differentials of cooled vs heated liquid
https://www.youtube.com/watch?v=7NwxMyqUyJw
Have you ever wondered how the refrigerators in your home work? Refrigerators, which have become an integral part of every household, work based on some simple and interesting scientific principles. Beginning with a basic refrigerator model, this video will elaborate on the operation of modern refrigerators, along with the secrets behind their h...
since the pump would need to be powerful enough to push the liquid through the throttle
Something something Spherical chickens in a vacuum
I try to take my reconstruction of the "Blow-Lamp" as far as I can, doing my best to reproduce the capabilities of the tool as it was described in T.P. Danger's 1830's glassworking manual; 'The Art of Glassblowing'
References(in order of appearance)
The 'Mutus Liber'(containing depiction of alchemists using blow lamps):
https://www.loc.gov/item...
Want to know the materials with the highest melting point? In this article, we will introduce top 10 materials with the highest melting point in the world.
Seeing this without any visible context is terrifying
Did we find space for the death beam
And old, but very interesting, movie from our predecessor Philips Cryogenics about the development and technology of the Stirling engine / cryogenerator. Part 1 (of 2)
Part 2: http://www.youtube.com/watch?v=GFfMruoRMGo
there should be space. This video demonstrates that with air flow, the direction of heat transfer can be focussed into one area, which might help our efficiency if the airflow can be achieved without wasting energy - for example by capturing wind
However i'm not sure if this makes any difference to just containing the entire assembly in a furnace
does the increased oxygen add energy to the system, or does it simply turn the existing potential energy into heat energy faster?
I'm not sure if my joke was completely mistranslated or i should be even more afraid

ok it turns out that the more air we feed into the burn chamber, the more efficient the burn. It doesn't add energy but it does let us use more of the available
Honestly, be afraid.
Do realise that if this actually works
Militaries would actually like this..
And whilst we could say no as much as possible, the fact is, it could be done which means money gets spent on developing their own ones...
If this ever does become real, it is only a Matter of time before humanity sticks a weapon on it, and that's just how it is sadly
the intensified flame can be directed onto the heated piston so that we lose less of it to the environment
Our little neuro going off to war...
I'm sorry for the other side
we should be aiming for very little fuel but high amounts of oxygen in order to have the best burn efficiency for the sterling engine. We can probably maximize efficiency by shrouding the hot piston in a graphite shell that's heated by a burning ethanol torch
Wdym off to war...
She's the general
It's us being shipped 
Do remember that she needs to be cool enough on the exterior to interact with people without first degree burns
Or for that matter, pretty much anything else
What's the human temperature for touching things? It's canny hot, but like, not fridged cold
fr we're going to have to pass the watercooling system along the exterior of the burn chamber
I feel like we can get away with a maximum of 110 degrees exterior before it she becomes unconfortable to the touch
We're not like 37ºC? Or I'm missing something 
Please mention the units you are using btw
You have British people here
Right... Yeah, as 110 c.. man that's something else xD
ideally the water cooling keeps her perpetually at a nice human body temperature
Give or take I guess
But also we have to take into account when she's on load and how hot it'll go, or if on load that's how much it gets up too
Well... what do humans do to coll themselves in hot summer? time to go shopping 
The neuro must absolutely be huggable
The question is how many gallons of water neuro is going to pour on her and how it'll affect her systems xD
She does have to be rainproof
Oh, I was thinking more of like, going inside a shopping center where the air conditioner is at full blast on summer 
As long as there's no gaps in the skin she can probably swim.. right?
very true, i almost forgot that
we need to figure out how to keep her waterproof
The swimming in question...
She might be able to walk on the sea bed, but swim?? Definitely not xD
Atri levels of enginiering 
Humans have buoyancy
Air that let's us float
Errr a robot is a massive hunk of metal..
I mean boats can swim but she ain't a boat xD
Boats float based on how much water they displace
Cute
🔥 looks highly flammable
it'd probably get in anyway
maybe we seal the joints?
what fabrics are hydrophobic
This might be a good start with the references but it might exist as a thing
Durable water repellent, or DWR, is a coating added to fabrics at the factory to make them water-resistant (hydrophobic). Most factory-applied treatments are fluoropolymer based; these applications are quite thin and not always effective. Durable water repellents are commonly used in conjunction with waterproof breathable fabrics such as Gore-Te...
Well she shouldn’t be setting cloths on fire when she wears them. She’s going to have to wear cloths. 
Well she probably won't be breaking any laws without any it'd just be weird
Aye but I was thinking that if she's in water and it sets off the electrics or fire, then that too is going off in flames as well xDD
The laws of man dont apply!
Yeah... and neuro is probably the first one that wants to buy docens of outfits for herself
Yeah she’ll definitely want to buy clothes lmao
The laws of men don't apply.
But the will of neuro is such she'd be getting clothing whether we like it or not and she'd own that clothing xD
She'd complain ;-;
We probably need something that's easily replacable. Unfortunately, we probably cannot create hair that grows so, every time she wants a new haircut that requires more hair than she has... (she'd need a new wig)
A wig that won’t come off easily.
Do remember, hair is heat absorbing, which is anti to cooling
Well the head wont have all that much going on in it right
Hair and a hood?
Plus we can probably make some synthetic hair that works as a heat sink or something
No need to project 
Ow my single braincell
She'd want hats for sure
Yeah like a frog hat lol
The thing with wigs is that the hair will get everywhere, so how are you gonna make it so hair doesn't get into body, is probably how one will make it waterproof
hair loss
I was thinking metal hair..
This single cylinder thermo acoustic engine operates from a tiny methylated spirits (denatured alcohol) flame. The optimum running speed is 2700rpm when warmed up fully. As far as we know this is the fastest thermo-acoustic engine available.
The thermo acoustic engine is known by several names, including resonant engine, lag engine, lamina flo...
Horror movie sound effects while brushing
Like, braids of tendrils of slightly thick metal hair
I'm not sure why I had that thought
Go to https://establishedtitles.com/THEACTIONLAB and help support the channel. They are now running a massive sale, plus 10% off on any purchase with code THEACTIONLAB. Thanks to Established Titles for sponsoring this video!
I show you how a thermoacoustic engine works to turn heat into sound that then moves a piston
Shop the Action Lab Scienc...
Ears
They use wigs on humanoid robots already
What?? 
It's not that it's not possible, it's just hair loss and hair entanglement inside the machine is my concern
They never had to deal with a neuro
Huh?
Wigs, not a misspell
Gotta make one for vedal that will turn neuro's complaining into energy

Honestly, not a bad thing to do is to capture some of her energy when speaking back into something else
Don't think it's worth it, sound doesn't take that much energy (or does it?)
Plus might muddle her speech
Humans voice get muddled by the mouth, what you hear isn't what's heard by others
Depends on how much space is used I guess
Idkkk I can't describe things atm
Space, development time, cost, hooking it into the network
The Chinese Academy of Sciences build a thermoacoustic engine which delivered a groundbreaking 140 hp (102 kilowatts) of power from a heat source of 986 degrees Fahrenheit (530 degrees Celsius). This is the first time that this type of breakthrough generator has gone over 134 hp (100 kilowatts).
Maybe something like
Her ear microphones would be feeding into a data thing which turns it into information neuro could use, perhaps an acoustic engine could help them run a little less costly when theirs constantly noise around?
how so?
I'm pretty sure the acoustic engines need a lot of noise to really work
Speaker placed on one side to work as a refridgerator levels of noise
this version functions via heat
the heat creates resonance that is converted into motion
Could also filter out backround noises from louder noises nearby to help her not get overwhelmed in noisy environments
it's more efficient becuase there's no moving parts, and therefore no friction
well
less moving parts
Well, there's material stress, but it can be pretty good. Just not 'turn noise in a crowd into useful energy' levels of good
the chinese submarine puts their alternator directly on the end of it so that there's no piston or wheel
instead they just have a copper rod moving back and forth through a magnetic field
Ah, yep, that's the 'internal stroke stirling generator' variant. I think it's quieter and more efficient
yeah really compact too
It requires making a linear motor though instead of a normal motor
we should go with that imo
okey, so... we gotta make one of those engines?
seems like our best option rn
no moving parts whatsoever, just air
so it's very efficient
question is if we can make or find a linear alternator
i searched them up and none are easily available
I mean, when we want an engine. I don't think having engines at all are MVP though.
depends what we want imo
They are pretty cool and have a lot of benefits, but it's more than necessary for getting a robot moving. So whether we do it dep-- yep, it's if we want to
24/7 running neuro would need an engine
short stream neuro wouldn't need one technically
What's MVP in this context?
i think neuro who does irl collabs would need an engine of some kind
it depends on what we want from her
minimum viable product (robot)
Probably. I mean.. neuro has to travel to america to collab with toma
A mechanically powered flashlight (UK: mechanically powered torch) is a flashlight that is powered by electricity generated by the muscle power of the user, so it does not need replacement of batteries, or recharging from an electrical source. There are several types which use different operating mechanisms. They use different motions to generat...
wikepedia pointed me here
these use linear alternators (sometimes)
oh, wait. CRAP! I used to have a radio that had a crank like that!
Wait, there's another type of crank...
I have one kf those
Those ones. Oh. Nostalgic!
loved these as a kid
OH god! I used to have one of those. I swear, they never worked for me
lol
i wonder how efficient the generator is?
if we can find where they source their alternators, we can make a miniature generator that's very space efficient
They probably make them in house
i don't know
That explains things
A flashlight is only useful if it has power to operate. No battery or a dead battery can be more than a small hassle when fumbling about in the dark. In order to ensure that there's always light available, you can create your own flashlight that's quickly powered with only a few shakes of the flashlight body.
I don't think those generators are super efficient. 
you might be right
we'll have to look into what makes an efficient alternator design
i don't think they sell liniar alternators commercially
Huh... I suppose we could just use multiple coils of magnet wire, or use a wire winder, then it would be a lot more efficient
The other option is to do this, but now i remember the target was to do "no moving parts" so it kinda defeats the purpose 
yeah that'd be the piston design
very true. i guess a lot of it comes down to how well we weave the wires
i'll look up some youtube tutorials on it
I feel like it'd be similar to making tesla coils actually, at least part of it
With tesla coils, you have to wind one side of a giant air transformer coil. In this case you'd need to wind a tube or a bunch of tube's instead
true
Testing the concept of using a linear generator to light up sea buoys to aid sea navigation. The jig is designed to simulate wave motion. The 12 bright LEDs did light up but will slowly dim when the wave motion stops. Done in 2010.
tightly packed wires seems like the way to go
i don't know how to make electronics safely
Avoid inductive kicks and charged capacitors
You might want a flyback diode somewhere around the linear generator so that if something disconnects on the circuits attached to the generator somehow it doesn't cause a huge spark

do you think a very long neodymium magnet and really long and thick coil would be more effective? Or is it better to have lots of coil, but small magnet?
I honestly don't know
seems like it sort of
we can probably surround the air piston with the mechanism about half way along
that might interfere with the generation though
i think fitting it on the end is best
lets say roughly 5cm long alternator for now, and a large and wide piston so that it has thick walls for handling the heat and pressure
so roughly about 5cm by 10cm
Sounds like enough space
Do you think fighting gravity by placing it upright would reduce the efficiency?
Hmm, I don't think so, usually there'd be some sort of spring to keep the piston in place anyway, and gravity would slightly displace the rest position
It might decrease the efficiency if the full stroke is outside the generator length, but I think that could be minimized
Nice, ty for the feedback/info so far. When we design this we can probably make a version that replaces engine space with battery space as well.
oh i assumed it'd be the opposite
becasue i assume the coils would be on the other side of the generator housing from the magnet
Ah, the magnets need to be as close to the copper coils as possible
ah yeah we'd want the magnets outside the generator then
we have the room for it at least
also we can probably make the generator itself out of ceramics
assuming repeated local heating doesn't cause stress
Er, the magnets should be attached to the moving part of the piston, insulated by some materials, and they should be surrounded by stationary copper coils at all times during movement
Ah, like this:
yeah i was thinking it'd be a rod in the ceramic housing with hollow neodynium magnets half way down the rod surrounded by very close wire coils
(Also, we'd want to double check how the magnets handle temperature)
that way the coils are as close as possible
very true
Ah, ok, I think we might've been in agreement the whole time then
This explains why on the chinese device, they went with a copper plate surrounded by magnets instead of the inverse
the rod would have to seperate the magnets from the housing ig
makes it longer but that's not too bad
Fair. Depending on the magnets that might be the best case even if that means the electrical contacts move
i think i might try to make a mockup of this kind of engine myself using the DIY guide for the shake-recharging flashlights and a basic piston.
not sure where i'll get the magnet from but it's probably available
Ok here's an idea, what if it was double sided?
Wait... that made me think: wouldn't the magnets be on the cold side?
wires in the middle, shaft on either side, engine housing around shaft
yeah
there would be wadidng in the way which i didn't think about as well
in this example the magnet would be the only part, so it'd be touching the heat, however, with this design there is a heat-seperator
lemme get a good pic
the part where the steel fastens to the tube is the border between cold and hot
Ah, I found a couple:
yup the latter one especially
probably for the best, since they won't ever get hot enough to become weaker
it makes the whole thing more simple
yea
i think if it were double sided, it'd work even better
we can probably slim down the large central section a lot and also the pistons themselves
probably. They might've been increased in size for the diagram
maybe not the start position, but it gives an idea of what it might operate like
the walls are thick in this model because we'd use ceramics and heat up the ends with lean ethanol and a fan
you can melt glass with those temperatures
the balls inside the tube are the wadding
Wait, why does it need two heated ends?
pushes the bar back and forth
cool area further towards the center ensures they both operate normally
it'd need two ethanol fires though which is a big drawback
There's probably a reason there are no stirling engines made that way
It looks more like the cold and hot chambers are enough to create motion by themselves, and yeah I think you'd have to synchronize it perfectly for constructive interference instead of destructive interference, so it could actually randomly cancel out if that's not handled
Alright, I think this is good enough for today:
I like your funny words magic man
A stirling engine is impractical. A humanoid robot requires fast and precise movements, which would be be easier to achieve with electric actuators. Stirling engines dont provide the immediate high torque needed for that. You could use it to generate electricity for batteries but it would still suffer from a slow start up and it would be inefficient for dynamic tasks. If you want to create a quadruped that moves slowly, it could work, but not for a humanoid robot.
It would also be too bulky and balance is important for a humanoid robot.
Ah, yeah, that's the plan: using it to generate electricity for the batteries
There are different sizes, some really small. It might be tough getting the right size considering everything needed for heating/cooling as well though
It would still be inefficient, all the cooling required could add too much weight, and it would still not work well for fast dynamic movements.
It would work exactly the same for fast dynamic movements because it's not planned to be directly attached to the motors at all, just charging batteries.
As for inefficiencies, Stirling engines can be quite efficient for engines, especially if made of high temperature materials.
The cooling is a good point though. Getting a good heat difference would probably require cooling equipment about the size of the engine... but again the engine can be shrunk down.
The real problem though is feature creep. There's no real need for the engine if we have batteries, though for Stirling engines there is a need for batteries due to the slow startup. It does improve how long the robot can move without recharging, but I feel like it's better to get some simple 1.0 version working before making all the improvements.
Also unlike the face where MatthewBadger seemed to have some detailed plans, I feel like we're switching from engine to engine without anyone really being an expert on the engines and wanting to implement it themselves
(Though, if there is someone, then it would be pretty cool to add)
We probably need a way to expel waste from the engines other than the human biological way
Or at least make sure it's non lethal
Yeah that's how I feel honestly.
I know most people and companies stick with Lithium batteries when working with humanoid robots, and they do it for a reason.
Jumping to try something experimental without expertise would most likely end up leading to failure.
Or worse, end up causing a fire.
I agree with the problem being feature creep. Right now our priority should be on the MVP (something I think we should formally define some time soon)
While on one hand, water cooling is neat and effective and all, but I can't help but feel weary of it. In my experience, hoses tend to not like being jostled around willynilly, and the places where water cooling is used is usually stationary ones; nuke plants, PCs, vehicles: Places where the actual tubing are kept stationary.
For this, the tubes would be almost always in dymanic motion, and if the vibrations from a car's engine is enough to loosen the clamps on the radiator's hoses, then the risk is high that repeated, strong shaking from the robot's movements would just absolutely push the clamps and hoses to their limits. Even if assuming that the hoses are kept exclusively to the torso - so no water cooling for the motors in the limbs - then that's still a lot of movement as the torso moves a lot when we move, and will be moreso for the robot.
Still, I'm open to it being at least attemped before writing it off. Don't know what works until it's tried in practice.
Also friendly reminder; feature creep is a slow and insidious project killer.
Additionally, the fact that the hoses in a PC can fail despite the most movement it'd get is from the fan and if it's an unlucky one, the user if stationed the desk, is a red flag for me, personally.
Can be summed up to manufacturer error, part failure or simply user error if they either built the PC themselves and didn't secure the clamps good enough, but the fact it can happen regardless and the aftermatch of such a failure tend to be pricy given, ya know, electronics now drowned with coolant, which is hopefully just water.
hose would need to be secured in such a way that full range of motion of the limb does not transmit force to the connection of the hose to the mechanical components. A tall order.
i can assemble a plan for only batteries but i absolutely plan on making an engined version at some point afterwards. Maybe v2?
the recent discoveries about limb locking mean that a lot of the reasons i avoided 100% battery power are now void
I think the MVP will be just getting it to move with a humanlike gait, be able to operate for at least 2 hours at a time, and mostly pass the uncanny valley effect
That's like bare minimum for this
Oh and a speaker in the mouth
pretty nice description for a minimum viable product
Skin is a "it'd be really nice to have this"
But probably also somewhat necessary for uncanny valley
The MVP for my next robot is just: "neck"
Getting limbs/a neck that moves right and can lift things is actually pretty hard, so that will be the work for my next wheeled robot arm
(Though, I have a bunch of under-powered motors and I'm making gearboxes for them so I can make more robots at near zero cost)
After we get all that and make sure it won't spontaneously combust then we can probably find a way to tell vedal about this
The MVP is all you guys 
true
Right now the only thing I know of for communicating with the robot from the AI is my kinda hastily made ad hoc code. It doesn't have anything for text or wifi communication, just camera and mic/speaker, but I could add different methods of communication.
If there was a screen, maybe we could output text/images.
im talking on discord
Oh, lol, the discord bot NeuroBot
¯_(ツ)_/¯
so you dont know?
she also reacts to pet
Huh, also some translation
i wonder if she will be communicating normally on discord
although
i feel like
vedal wouldnt like that
mostly cuz he wouldnt have much controll over said conversations
might be wrong tho
¯_(ツ)_/¯
He mentioned potentially letting her admin the discord in a serious talk with her while on Scribblenauts. I feel like he would once she’s mature enough, more intelligent, and less likely to do what people ask because it would be funny. She’s not there yet but maybe someday. Only Vedal knows. 
i see
when was that?
There’s a clip of it, one of their deep talks during the subathon while they were playing Scribblenauts.
I feel like 'more intelligent' practically just requires more parameters, which means more RAM/VRAM.
Maybe some sparsity could improve that with really wide networks, but I haven't seen much in AI showing that it's possible to get more intelligence out of a same size transformer neural net without a lot of training, more parameters, or recurrence which limits training to sequential stuff.
well neuro's life is pretty much one big training
Facebook's llama should have had quite a lot more training than Neuro
how long is llama in usage
I don’t want to speculate too much on how he will achieve it. It would very much be kept a secret once he does, but she’ll most likely get to that point eventually.
The ability to think rationally as well.
Even in biology, more intelligence is pretty strongly correlated with 'more parameters' or more neurons: https://en.wikipedia.org/wiki/List_of_animals_by_number_of_neurons
There are definitely improvements beyond that, both in AI and biology, but at a certain point, you just need more RAM
I'm not sure the requirement is necessarily intelligence though. It might be more security, where random people can't just say the wrong thing and have Neuro delete all messages or something.
He wants her to be much more intelligent.
Well, maybe I could share some sparsity kernels and sparse backpropogation methods. That would probably help a lot
He knows what he’s doing lmao
Just let him cook
Eh, the more the merrier
you could ask him if he wants to
I still need to fully integrate my stuff into pytorch though. Right now it's on a mix of OpenMP and Vulkan. Translation from vulkan kompute to cuda should be something most LLMs can do though, it's the other way that's harder.
Yeah just don’t ping him lol
well thats obvious
thats like
going nuclear
on yourslef
lol, yea
Alright. I’m going to get more hours of sleep in.
Y'know, I think 'sleep' is one of the things AI/Neuro could use to really improve memory/intelligence.
In REM sleep, people play back memories but also see how things could have been if they acted differently.
Maybe if the top error states were recorded and played back, that could work as AI 'REM sleep'.
It might be possible to store some sparse activation states in cpu ram/hard drive during the day, rank them by highest error, and play those back when offline.
Edit: and inhibit the activated attention states so other parts can be trained.
I think there's a huge 'should've thought of that' problem with cross product attention you don't have with normal neural networks that the 'REM sleep' would solve
i agree
Ah, found the vid explaining why I think this: https://youtu.be/SRfJQews1AU?t=639
Grokking, or the sudden generalization by AI models to new knowledge - that occurs after prolonged overfitting in LLMs, is a surprising phenomenon that has challenged our understanding of deep learning and AI in general.
While a lot of progress has been made in understanding grokking, finally we get some answers -we have been waiting for 7 mon...
Backprop becomes zero sometimes in the attention mechanism except for the activated portions, so minimal learning happens. There are a lot of ways to deal with it, but they kind of feel like hacks to me, like adding randomness or smoothing things.
Wow, I did not expect TPU to be way better for gears due to its flexibility.
Apparently it can avoid pitting/wear really well because it bends/squishes, so it lasts a lot longer, while the ultimate tensile strength and yield strength are still both quite high
ABS got a red ~50% safety factor for pitting, while TPU didn't
Then again, raw aluminum and aluminum alloys... my calculator might be a bit off (though 1mm module is really small)
Honestly, maybe just coat the gears a little if we really need high torque & small size. Though that could increase friction... unless it's something like PTFE.
ABS coated in TPU or pure TPU may be able to perform 50% better than pure aluminum gears. For pure TPU though, helix angles or extra contacts may be necessary so teeth don't flex. TPU can be relatively low friction though.
Helix gears have an axial force though, which needs to be countered if it's not a double helix. Double helixes require print-in-place for planetary gearboxes.
my irrational dislike of tpu defeated once again
heh
practise practise practise
I see you're practicing to get a head
haha
i did some research for v2, and a stirling engine is the wrong way to go unless we use an autoclave to increase pressure in the system, as we're losing as high as 99% of the available power otherwise.
(i did this research because it interested me)
gonna work on more blender practise later today
Question i thought of while trying to sleep: do the human strength calculations include adrenaline doing what it does or just average human strength
Huh. I was thinking a burn chamber to increase temperature due to engine efficiency. Stirling engines sometimes have special gas inside though. Where did you find we need to increase pressure?
As the pressure in the chamber increases, the force exerted on the gas by the same amount of heat also increases. This pushes the piston with more force which results in greater power outputs.
And old, but very interesting, movie from our predecessor Philips Cryogenics about the development and technology of the Stirling engine / cryogenerator. Part 1 (of 2)
Part 2: http://www.youtube.com/watch?v=GFfMruoRMGo
Oh! The efficiency remains the same actually. I think the power output rises because it has to heat up more air
without high pressures, stirling engines woefully underproduce compared to conventional engines for their size, although have higher efficiency
yeah
the power we get out of the system increases, but the efficinecy of the fuel burn is about the same i beleive
we can't pull more energy out of the fuel than actually exists
30-60 atm is tire pressure though. Seems fairly doable for a sealed chamber, though long term you'd want something to maintain internal engine pressure
a safe pressure would be ideal so that it doesn't become too dangerous. This is why i think an autoclave housing a Stirling or thermoacoustic heat engine would be ideal
yup
lemme grab some pics
Wait, no, tire pressure is 30-60 psi. It's 2-4 atm.
Still doable, but now it's harder
autoclaves like these are designed to be heated at high pressures; it's rated for about 400 degrees celsius with a working pressure up to or below 45MPa (444 atmospheres)
the hollow tube within would be ideal to inclose the wire spool, magnet, and piston
a contained design is ideal for higher pressures because it doesn't rely on rubber diaphrams. however for lower pressures, rubber diaphrams can work as well
higher pressure designs would require that the wire exit the autoclave via a high pressure valve or seal
my findings show that unless we went with a pressurized system, the stirling isn't a good option despite its low maintenance and use of ethanol, so therefore if we want to use one, we'd house it in a pressure vessel like this for maximum safety, even if our preference was to go for a lower pressure design
i personally think that a presurized stirling engine would be a good option
but only if we can use it safely
it's also quite costly at around £1200 for an autoclave of sufficent size
0.5l Mini Pressure Vessel Reactor Autoclave - Buy High Pressure Reactor
pressure Reactor
mini Pressure Vessel
autoclave
pressure Vessel
mini Reactor
500ml Reactor
100ml Reactor
250ml Reactor Product on Alibaba.com
500ml is about this volume on the inside
enough room for the alternator + piston
Also, going with a pressurized gas would allow us to use a much more effective gas type
Yep. Iirc Helium transfers heat much better
yup
Seems like designing one would require finding the optimal tradeoff between total volume vs pressure
There is also an improvement that i came up with, which is to attach rods to the rear of the piston head that loop around the piston chamber and intersect the cold-side entrance of the regenerator. Attached to the rods are choke valves sized to the inner diameter of the pipe. At the beginning of the forward stroke, when the gas is cooled, the choke valve is forced into position within the pipe, fully closing at the end of the stroke and heavily reducing the pipe diameter.
As the pressure of the heated chamber builds, the gas is forced into the regenerator and pushes the piston simultaneously, forcing its way through the narrow hole of the choke valve at the same time, which cools the gas. As the piston is forced back, the choke valve is opened, increasing the diameter of the regenerator again.
This difference in width on the forward and return strokes ensures that the gas is cooled on the forward stroke, but not cooled on the return stroke.
yes that's true
Needs physics/mechanical equations to verify... I'm not up to that rn
i'll build one and see if it makes a differrence in efficiency
my biggest concern is whether the gas needs to be condensed in some way artificially on the return stroke in order for this to work properly
might take me a while to get the materials together but it'd be a fun experiment
Yeah, I imagine actually building some of these could be hard, especially the optimized ones
will probably need to buy a torch for silver welds, some steel tubing of various diameters, some steel rods, a closed tube, steel sheet, and a bandsaw
oh and some porous heat-capturing material for the regenerator
maybe a thin drill bit as well and a drill
oh and something to check rpm with
if the choke valve design works out, it should produce higher RPM than the regular design for the same amount of heat
for that reason maybe a thermal gauge
might get more output then, though also might lower efficiency
huh true, i guess efficiency can only be measured using a linear alternator
Yep, power out / power in
yup
i guess that's an excuse to start learning how to build those
that way if the test bares fruit, i can make a proper design after
neat!
(Idea: I wonder if it's be possible to attach one of those 3D stable diffusion models to neuro, like have her LLM act as the transformer in the diffusion model before going back to being an LLM, and then she could slowly generate video game worlds as if she were in a lucid dream... might need some vision and other additions though)
That would probably increase latency.
Latency mentioned
Definitely. I plan on making a stable-diffusion based AI for myself rather than an LLM one though (so, speaking is generating Fourier transforms for example, moving is generating poses, etc.), and that's going to need a lot of improvements to get to something reasonable like 30fps.
It'd only need one step per fps rather than 20 though, since it'd be working off the previous frame instead of noise, but I think that still needs a 10x-20x speedup or so...
...maybe I could get away with 10-20 more GPUs for it...
Maybe find a bimetallic thermometer to hookup to a thermal exhaust opening that's closed when not in use?
You can use a soldering iron for some types of silver welding. Depending on if it has any iron content in the alloy of filler metal. Your cheapest option for welding would be an oxygen acetylene torch. For a better joint you would want to use Tungsten Inert Gas welding with a filler rod.
A warning when it comes to sheet metal. That's a whole other can of worms compared to just welding and cutting other thicker steels. You need some specialized equipment to do certain bending and cutting operations. Those operations can also require some extra math to account for springback and stretching of the metal. A good resource for that would be the Machinery's Handbook.
ty. I'll have a look at it and incorporate those methods when i make the low pressure choke valve prototype
This thread, but engines:
yamaha be like:
Combustion engines.
RC plane engines.
Petrol engines.
Glow fuel engines.
Ethanol engines.
Stirling engines.
Crank lantern generators.
Shake lantetn generators.
Stirling acoustic engines.
Preasurized Stirling engines.
I wonder what would be the next idea
blood run engine?
wait is human body an engine run indirectly by blood?
No, not really, oxygen is transferred around using blood, but not the ATP that's required for energy
I would see it more like an engine cooled by blood 
Which is the last step needed in the ignition
the blood is the part that runs
accidentally reinventing ultrakill
i actually forgot about ultrakill when thinking about this
so accidentally is correct here
Like oxygen's place in the ATP cycle is very important, it's basically the without this it doesn't work, but that's the same for the rest of it.
at the very end of the ATP cycle one oxygen is added which is given by blood, but the start is microcondira
Microcondira is not named the powerhouse of the cell without a reason xD
but like
do u guys think making an engine that runs on blood is even possible?
in theory atleast?
I mean, blood is edible, so probably, but it's also probably not a good idea, for obvious reasons
The problem is... blood is.. organic. Where do we get blood? Normally, we produce blood 
Vampire Neuro
well
blood banks?
wait
blood banks are a thing right?
why blood being edible is an requirement? i dont think gas is edible
It means we can turn it into energy, so it's possible to turn it into energy
Nah. Oxygen isn't flammable
Yikes
oxygen makes other things flammable instead
i see
This does give me some engine-related ideas though. In pure oxygen, fuels burn much more easily even if there's water around them. Moisture content around 50-70% is hard to combust in pure oxygen, while 20-30% is hard to combust in our atmosphere, so a pure oxygen chamber would have a much easier time burning fuels... that would be a pretty advanced addition to any engine though.
we just need to figure out how to burn water
it's made of hydrogen and oxygen, two fairly burny things
ah this gave me an nostalgic memory
wait lemme bring up what i mean
That's hydrogen fuel from water electrolysis. You need a graphene lined tank to store the hydrogen, and then it can act as a pretty good battery
we just need to shove a electrolysis whatsit into the robot and a hydrogen generator
I think the graphene part's beyond current material's science or manufacturing, so we'll just have to advance those fields
not only would it be advanced, it'd also be incredibly dangerous
a cool idea though
True. Pure oxygen is pretty dangerous, especially if heated
(Side note, I might take a break from gears a bit to look at this stuff...)
https://unsloth.ai/blog/deepseekr1-dynamic
https://www.reddit.com/r/LocalLLaMA/comments/1iczucy/running_deepseek_r1_iq2xxs_200gb_from_ssd/
https://www.reddit.com/r/LocalLLaMA/comments/1idseqb/deepseek_r1_671b_over_2_toksec_without_gpu_on/
(If that's all correct and my sparse stuff works with it, I might be able to train models a bit smaller than that size in real time with 10-20 or more cheap RAID hard drives, since SSDs break if you re-write too much.)
sounds awesome
not too caught up on the deepseek news but i heard it was good
might make a good tool for navigation and proprioception
Navigation maybe. Procioception is more fast response. Running very large models off of hard drives is more useful for slow and complicated processes, while GPUs will be much useful for simpler processes that require fast reactions.
true
Pure oxygen is a poison
my bad i was thinking of liquid oxygen
Depends on the environment I guess. Hyperbaric Oxygen Therapy can definitely be bad for you. I've also known people who huffed oxygen gas from a tank in a welding shop (as a joke) and were perfectly fine afterwards.
@ruby lagoon
How many planet gears will your gearbox have?
It seems like 3 or 4 works pretty well for finding most gear ratios
...honestly, maybe I should allow a range of planets to the gearbox UI
No sweat I'm just trying to understand your calculator. Did you build from scratch using formulas or is it a series of modules in freecad auto-finding the forces and stuff?
I built it from scratch using formulas
It requires the freecad.gears plugin to generate the 3d gears, but right now I'm getting the UI and calculations right separate from freecad.
Freecad doesn't really have anything to quickly find the forces. I might be able to use the Finite Element Analysis plugin after making 3d gears, but it would be really, really slow
RIP. If you want any help with Force Analysis I could try.
Although FEA is WAY beyond me.
Fair. You can look at the code if you want
Did you write that, if so that's awesome.
Hah, no. I'm not that into gear stuff
Would be awesome though
It was an example of the different papers I was looking at for the code
Ah
i got sidetracked by plasma toroid fusors and acoustic levitation 😭
We're just doing technobable now ok

Wait no one here DM'd Ellie did they? (jk): https://youtu.be/LGMNlM4yL7A?si=dxGA04MCgvij2y4f
Ellie Gets Weird DMs About The NeuroDog...
If you enjoyed watching my edit, please, consider subscribing to my channel and leaving a like! Thank you!
Showing the support would mean everything for me! That way I can be able to keep improving my edits, so you guys could enjoy them more!
Ellie on social medias! Please support her as much as you c...
I have a hard time pinging moderators for important stuff I don't think i would survive pinging someone with more than 100 viewers
@floral quarry you have been drafted into the neurobot development team
Resistance is futile
Wait, based anti big tech dev?
Even if he's not actually a dev it's still funny to draft people for the cause
fair
Anyway back to imaging a neruo live concert
Yes gear design is covered to some extent.
what are we making
remember that response will be limited as I have exams
and all blueprints shall be done on a drafting table so do not expect cad models
I have a potato laptop which I compensate with my drawing and a slide rule
An android.
sign me up
I'm bored anyways
Ok as far as I have read we are struggling with power
And to that end y'all have come up with an engine
If you folks are serious about following through with this, then I suggest someone start a google doc or a shared document of a sort
I was working on stuff with artificial muscle fibres with one purpose: to speed up movement
That was until I reached end of school year and abandoned it due to the million exams for college entrance
There are two ways we can go: either choose muscle fibres or go fully mechanical
https://youtu.be/fBNX-OXdolM?si=Yna2vXE1rdkR843v
Something for a small battery setup
Here's the first prototype for my custom armored suit that is under works! I'm creating a real life suit. Not Batman of course as his suit has many impracticalities -- it's a Comic Book after all. So here is my in-progress tactical suit with carefully designed gear that would make sense in real-life, while retaining themes and styling to make a ...
As for engines... you folks are certainly going the right way... but some things I have read make me feel thermodynamics would not play nice
I would have immediately suggested a fission reactor was it not for the fact it is illegal to even touch anything nuclear in my country
Not even the ores
Be warned: if you want serious power, you will eventually need a powerful engine
We already have a doc tho
What country are you from
We're not necessarily making an engine first. People are just obsessed. 18650s and 21700s are good enough to get things running and I have a UPS circuit for them. (Engines are better for power / long term running, but harder)
As for muscle fibers, I made a gearbox designer ui for planetaries, and I'm doing split rings after, so we can have powerful pulleys from motors ( or Series Elastic Actuators)
Google docs are pinned
@floral quarry ty for the feedback
Wait, I think we still don't really have a design doc. We have a research doc and a history of work doc, but nothing like an organized agreement between all our plans and priorities. It's not clear what the first robot version should actually be and what tech it should have, or the second
yes that's true
we've been making it up as we go along
i think we should make a document that defines what exactly our end product should look like
and then breaks that up into steps or avenues to achieve that
(Ah, idk what it's called otherwise, but I think 'video game design document' is kinda what we're missing, but robot design)
part of it is that we don't know what we will come up with that can be added, yet. We know that she needs to have arms, legs, head, torso, but not in what form that should take
we've settled on some things, but those things were come up with on the fly you know?
it's very 'in flux' at the moment
what do you think?
I mean, that's fine. The design doc can be modified as we go. It doesn't have to be set in stone
Mind if I add in some 'before humanoid' or 'before face' robots? I feel like we should have a few versions, like neurobot with and without an engine.
I feel like such prototypes would come before v1 and probably whilst ellie is working on her robot, if she chooses to.
We can give them a designation in the design doc
Neurobot-1000 design document Introduction: Neurobot-1000 is a project that started out as a thought-puzzle about the benefits and drawbacks of a Neuro robot that closely matches the appearance of her latest vtuber model (as of January 2025). This idea spiraled into a project to design an ex...
i'll give you access once you're in
Need to summon a mod to pin
Thanks
no problem 
On mobile rn so I won't be able to do much for a few hours, but sent the request
Wow, fast mod
new document
this is getting serious
Watching in the shadows
More docs than arg crew
We're gonna need them. We gotta put eventually schematics of neuro bot several times.
Iirc the main things we need to figure out for v1 is how the face works and how the legs will work
@ruby lagoon personally i'm thinking that any prototype versions we make should consist of parts connected to placeholder components; An arm connected to a desk, a torso connected to nothing, a head mounted to a fake robot or dummy, etc.
This way we don't have to spend time designing parts we will scrap in version 1, and can assemble the parts togther later.
So for example if we have completed legs and torso and arms, but need a placeholder head, it'd just be a camera mounted to a steel beam with gaps missing where head/neck motors would ideally be placed.
I'd make 3d models of these individual components as well as the v1 version. This way we are most efficient with our time and also, do not get a skewed perspective that an improperly designed part would create.
Engine as mentioned before is kinda optional
arms too probably. I'm thinking of v1 as minimum neuro that fulfills design document's requirements
v2 can exceed them or refine those things
I thought we had arms down pretty good
for example i'm fairly confident we can cram enough batteries into her to give her 2 hours life if she's conservative with movements. So v1 would be entirely battery-powered
oh you mean we gotta finish working on the things you listed?
that, i agree with
Ye
rn face is very up in the air. I'm thinking we stick with small brushless motors due to their precision and dexterity, which is needed for the face
a lot of humanity comes from slow, delicate movement
Face is definitely the hardest part
Fair. I was planning another wheeled robot arm for myself. That's a bit better than an arm connected to a desk. Maybe after that we could add a torso to get something like R.O.B., then humanoid with screen head like in Stray if the face isn't done after that.
Something i don't think we've talked about yet but how long is it going to take to charge all that with a standard cable
yeah that'd go hard
also it'd be hilarious
What if just a standalone leg
A single leg
A strange video apparently shows some legs jogging down a dark street without being attached to a body – A total mystery by conspiracy theorists.
love that movie
Lads i just realized we were missing the obvious:
https://www.diy.com/departments/invena-50cm-1-2-x-1-2-flexi-flexible-kitchen-basin-monobloc-tap-connector-hose-pipe/5907798716210_BQ.prd?&&&&msclkid=1bd504fd3d461943abdb6363755fefa9&gclid=1bd504fd3d461943abdb6363755fefa9&gclsrc=3p.ds
Those leak real bad at the ends
india
i'll check some vids of them in operation
born to fuck around, forced to find out