Ai Dreams Forum
Software & Hardware => General Hardware Talk => Topic started by: goaty on May 19, 2019, 02:30:22 pm
-
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/60704869_911844662496925_1564253387287953408_n.jpg?_nc_cat=111&_nc_ht=scontent.fper6-1.fna&oh=71617205500b968cdb49d55adea02eca&oe=5D5966D0)
This looks like the easiest way to get your muscles done, Its a simple attachment of "ropes" to a hollow inside, and you need elastic/bellows/bags to allow for the translation to rotation that's going to happen.
-
That's really good! I will copy if you don't mind..
-
Id love it if you did! :D Yes! it looks attractively simple! I like it too! Its in the "soft robots" class, of body designs. nifty.
I knew I was onto something when I splayed out the connections to the joints, seemed like the right way to do it.
But the system isn't quite finished. I just put it into the computer, and it looks like its not stable enough. and the diagram is missing an important element, it needs a round obstructor I think, so it obeys a rotation when you pull on the pulleys.
The "bags" are the equivalent of a ball joint (and just for trivia, they were used in old school cameras ages ago, but we are using them as a joint here .:))
and I don't think the tendons are going to able to hold it tought properly... ill be back if I come up with the solution.
It really needs someone to go craft it in real life then the problem&solution will be more obvious.
Sure looks easier than hydraulics doesn't it, even tho you can put water bags into it, and its basicly the same idea, except we just got rid of the water.
-
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/61082725_916354835379241_5537818324794081280_n.jpg?_nc_cat=107&_nc_ht=scontent.fper6-1.fna&oh=3a92911cc11504b4a13c731641bd8cb0&oe=5D99D4F6)
Hey HopeFullySomething->
I fixed my problem, you make it one string against a bearing that is on the rotated rigid next piece. so you pull either way, and the bag will stretch with the pully string.
If you were getting exorbitant, youd put a roller where the bearing is, but what if there isn't enough time? ;)
-
Cool, I like it! I'll just need to figure out how the springly part will function. Maybe it doesn't need to be springy, just flexible. Needs to be able to bend but at the same time resist contraction.
-
Be my guest if you want to try and implement it, remember theres no warranty if it ends up causing you misery. :2funny: btw, just use a tshirt or whatever for it, don't need no flex d00d.
Me myself lately im imagining a cool little thing made of voxels for it, make it extra easy... stiff voxel, cloth voxel, stiff voxel, cloth voxel, chuck some ropes in. potato potato ;0
-
Alright, my quadruped "Blinky" creature needs some sort of legs, and I've got a lot of spare shirts. This should be interesting. O0
-
Alright, my quadruped "Blinky" creature needs some sort of legs, and I've got a lot of spare shirts. This should be interesting. O0
Youd be a total star if you did.
The shirts have to support the weight of the bot, so maybe youd need a few layers, or the rope will rip them. Interesting to see what u come up with, but if u end up not wanting to bother, flake out I don't mind I should be the one proving my own designs after all.
-
I thought some more, and I think that you don't attach on the bellow, you attach just after it, and u need to get the length between it and the bracing right, to give it the play it needs. and the bags have to start off with a bit of compression on both sides so it adds up.
-
The main physics of the design are good. The wide structure resists breaking/bending and protects the moving parts inside it. I'll probably have to go through many many itterations to get the details right. If anything ends up working i'll post on my project thread.
-
np. I should be putting this together myself. just cut in so many directions for this bot im doing. physics engine, computer vision, body muscles(this) power source, input and output, then the computer, then manufacturing, and a senile idea to get infinite processing power, etc etc. I can only do one thing at a time, but usually just wasting time on youtube, just to make it all grind to a halt.
Nope, design not working, There needs to be an obstructor of some kind, to keep the joint always the same length no matter where its pointing, also springs/flex could be used here...
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/61584181_917176291963762_674753900665896960_n.jpg?_nc_cat=105&_nc_ht=scontent.fper6-1.fna&oh=117222e7fa09ac731a9d657cec815ca2&oe=5D5C59B1)
Heres my computer simulation of it, with a voxel creature maker - the magenta cubes are the elastic ones, and the cyan ones are the rigid, and they have pointy ends to touch the surface with, because I think this might make it easier to get going with at first. bit more to fix before I get to see it go, but theres a few things about physical walking that I have to work out yet...
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/61072752_917254715289253_5964617711852978176_o.jpg?_nc_cat=107&_nc_ht=scontent.fper6-1.fna&oh=de059872778828ac1cc2a74eacba82eb&oe=5D53D7A4)
-
Found it! Was trying to design this from memory and it wasn't quite right...
-
Cool! :D id feel the love if you prototyped something like this for me, cause I don't know if its going to work or not, its all just pictures in my head still.
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/69456250_979949805686410_2862064363789877248_n.jpg?_nc_cat=103&_nc_oc=AQmKlKsu9BRYK2wwNJYbTYEGnRBbAFJsRLkLd4ZW6sRmV5od3KhJ9BVoCU4SPrnZIR8&_nc_ht=scontent.fper6-1.fna&oh=0c45eed1daf5c15065a3fc73a46b9ba4&oe=5DCDF08E)</img>
Its not a complicated thing, you should put a spin on it, but how I see it - its all pretty much easy to understand except there needs to be an obstructor rod which is going to sit against a rotating plane to cause a rotation when it spins.
This design is cool to me, cause you could probably 1 hit cast it with whatever material and it could miniaturize really well even to the size of a 2c piece.
The thing not in the explanation here is of course how to add the linear drives which pull on the tendons, and how to pass the power from the battery through the computer to set their positions. You can use electric or pneumatic, and i'm such a dodgy bastard id say its better off without gears or spoolies, they just complicate matters.
Theres actually another roboter on the internet who made a beautiful pully wire worm, and he hates gears in em too :) you can join the club if you want.
-
https://pdfs.semanticscholar.org/3396/e17daf570f966ef7722446ea3e7ea7a52a75.pdf
:)
-
there needs to be an obstructor rod which is going to sit against a rotating plane to cause a rotation when it spins.
So it will have to hold the weight of the robot? Is it 3d like a ball and socket, or 2d like a hinge?
The thing not in the explanation here is of course how to add the linear drives which pull on the tendons, and how to pass the power from the battery through the computer to set their positions. You can use electric or pneumatic, and i'm such a dodgy bastard id say its better off without gears or spoolies, they just complicate matters.
The cheapest might be regular electric motors pulling string on a stick, like a puppeteer. Hehe I feel the same way about valves, I went out of my way to design a valveless steam engine one summer. It would need to be really big and heavy to work though.
Theres actually another roboter on the internet who made a beautiful pully wire worm, and he hates gears in em too :) you can join the club if you want.
Yeah, that type of thing is really fragile. (gears, valves, bearings) Needs to be exactly right or it stops working. Not the type of thing that can wander around on an alien planet for a hundred years, (as things should). :)
Something that has loose tolerances, which can take damage but keep working, that's how I envision the best machines.
-
Here's a small-scale prototype leg that I made some years ago.
https://vine.co/v/eUltL1B9ur6
I never got around to motorizing it; in the video, I'm pulling the tendons with my fingers.
One difference vs. what I think goaty wants is that the tendons aren't fully internal to the legs. I put them out on those little spars for better mechanical advantage.
-
AGI is a very serious pursuit. :)
(https://i.ibb.co/rtGTYfS/Screenshot-1306.png) (https://ibb.co/8BDTq4p)
Tried and discarded a bunch of designs, mostly they ended up being really complicated and difficult to construct. Finally got something simple, only two parts, housing and legs, and the legs are just the same part repeated. Still need something to hold the joints together, and a distribution for the 8 strings required to control each leg. Maybe just tech-flex air muscles in place of strings? There is probably enough room for that.
-
there needs to be an obstructor rod which is going to sit against a rotating plane to cause a rotation when it spins.
So it will have to hold the weight of the robot? Is it 3d like a ball and socket, or 2d like a hinge? (just think of it like a bone - he needs bones there just like us. :)
The thing not in the explanation here is of course how to add the linear drives which pull on the tendons, and how to pass the power from the battery through the computer to set their positions. You can use electric or pneumatic, and i'm such a dodgy bastard id say its better off without gears or spoolies, they just complicate matters.
The cheapest might be regular electric motors pulling string on a stick, like a puppeteer. Hehe I feel the same way about valves, I went out of my way to design a valveless steam engine one summer. It would need to be really big and heavy to work though.
Theres actually another roboter on the internet who made a beautiful pully wire worm, and he hates gears in em too :) you can join the club if you want.
Yeah, that type of thing is really fragile. (gears, valves, bearings) Needs to be exactly right or it stops working. Not the type of thing that can wander around on an alien planet for a hundred years, (as things should). :)
Something that has loose tolerances, which can take damage but keep working, that's how I envision the best machines.
Yeh that "obstructor rod" Is a ball hinge bearance on the end, if you want it to be.
That "lever" idea is cool - I didn't think of that, but wouldn't it take too much space up - because the lever would have to be inside, I would think a linear drive would take less space up. but the lever will count as a change up transmition, depending on the length, the same as a cog does.
But Hopefully Something, if you've got a better idea, doing it your way is probably better for you, considering I actually haven't even built it yet to know what its really like, it might actually be really bad, what ive come up with.
The main thing I was trying to achieve was to get something that I could caste in one piece with resin or rubber, so to make a virus all I would have to do is miniaturize the die.
I see you've got a program running?
Do you have a physics simulation for your robot? On that last picture? Record a video of it for me would be cool, id like to see your software!
Writer of Minds - that leg is really cool! Thanks for showing. Yes I haven't even built mine yet, its still just an idea, maybe it is weak I don't know. I like the return springs, im not doing that in my one, im having two separate pulling locations back and forth.
Because im not using a spring, I dont have to worry about getting the spring strength "just right", also it doesn't fight against itself - I just pull either way and its equal power.
-
I'm sorry there's not much physics that I could build into the program. It's just constrained to a 360 axis at this point. I'm not really sure where to go next. I've started walking a lot since I got a dog , and now I think bipeds are really cool, go figure. Maybe its the versatility outweighing speed and efficiency. Anyway, one thing I have figured out is the top part of the leg is a great space for air muscles. They would have to be doubled up, (8 of them), 4 connected to the body, 4 to the lower leg. That way the lower leg is lighter, and you don't have cables and air tubes everywhere. I'm not sure where to take this next (maybe our designs will diverge), but I think I'm going to incorporate springs at some point lol.
https://youtu.be/hUmSl7K-bSA
-
cool! 8)
If your joints aren't constrained properly, youll get unpredictable results from a genetic algorithm doing locomotion. then if the physics is wrong on top of that, it might end looking like its fighting against a strong magnetic force, and its no fault of the a.i. unit running it, its just its world and body is off kilter.
Crab -> Dog -> Biped, its in order of difficulty of moving around. (but once the crab grew a brain evolutionarily, its brain could definitely already be a biped if it had a body transfer, but I think when the creatures were all simple, the crab was the first form with legs (back when it was but an early plankton), because if it falls over, it usually is a safe fall it can recover from, not like a human, for example.) And Boston Dynamics are amazing... getting the dog to walk is more advanced than the crab is, just the way im looking at it right now. But could be wrong.
Air muscles I hear!!!! Sounds good! the guy that made the Strand-Beast definitely had the beautiful wind power in mind for his units.
You could use air muscles with the pully tendons - in a duo - but if it were purely air muscles, if you pushed on both left and right, the leg would expand out straight, and sucked back on both, the leg would detract like a snail. would be funny if you let the robot feel around to do it if it needed to. :)
Ive got a tip from thinking just today about wind power, but its only a speculative one, that if you connect syringes to each other, skinny to fattest, itll actually be like cogs, and if you had wind pumping the first syringe at a low psi, it might be able to push the fattest syringe at a higher psi - to try and compress into a battery to make it last longer, and give you more strength. But im not totally sure, I have to build the bastard to find out. It sorta fits in a box type shape in a sparse way, because it gets shorter every time it has to get fatter, and it just fits in a cube, so its not much space taken up - for the wind charga! But if it turns out im wrong, ill just have to be a bit sheepish later about typing this right now.
It only works well if volume counts as watts - then the thru coordinate counts as volts, and the radius coordinates count as amps. but I could definitely be wrong, as im only beginner at bringing maths to reality, not just staying inside the computer which is a lot easier and more can be simply supposed... in life it has to actually be the truth.
Heres a concept page of what im doin atm
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/70406835_997883817226342_7340853764682678272_n.jpg?_nc_cat=105&_nc_oc=AQlWXl7pUvLeeapcmPibGCbCKRthjcqroFKrTAfL9fxoh7DvsU0GNnVgtgoWW0fs2ms&_nc_ht=scontent.fper6-1.fna&oh=fe2bfb5c3f5769fcb1d286d737c60775&oe=5DF18B6E)
Air circuitry - 5 layers of lazer cut plastic sheetings?? (balloon powered.) - but its only going through keyframes, no brain yet, need to get the body plumbin' workin first, then I can get on with the amazing stuff later.
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/70655936_997885430559514_8022732440567021568_o.jpg?_nc_cat=105&_nc_oc=AQm9FQtC-Eo1YHwTGuawZp9ZcjnHXsy-stfRWbFfDRL8ZLKLD9Nj-PEobgunGoZcZrE&_nc_ht=scontent.fper6-1.fna&oh=9a5798c6362923661b8b2903de3d0c8c&oe=5DFF40B7)
Its a bit like explorers if it works.. but... the syringe to syringe transmission idea-> (definitely looks cool...)
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/71098151_997902703891120_1061967682259124224_o.jpg?_nc_cat=107&_nc_oc=AQkTL6MYoQNFR9P-MWyaNI9nBUBvO96NTP-mr2AThrxhWzYSYKkyJyCTWDCl-JIoT74&_nc_ht=scontent.fper6-1.fna&oh=d789ac740a92a1f9734aef4d5fd4f50b&oe=5E35B606)
-
Syringe to syringe... Soz dude... Nope... First law of thermodynamics and conservation of energy and all that... Stuff.
:)
-
If the power supply (air gathering device) has low force but constant action, then it would make sense to compress air with a small diameter syringe. Pressure = Force/Area , so if you've got a small force you can still get a big air pressure by using a small area to compress it. Then just store high pressure air for later use with "muscles" which have bigger surface area. Force = P*A, so any force is possible.
-
So im wrong about the syringe to syringe transmission?
I was thinking that if I had a skinny syringe, I would get a torque advantage over the fat syringe. (If you were "syringe wrestling" the skinny syringe would win) But I might have everything backwards..
You can hand pump up to 300 psi or something, I watched this guy do it, hand-pumping the whole day long, and with this thing you could get 3,000,000 psi just out of your arm. So, it may be a little too good to be true, so when I put it together, it'll just be a self demo of air tightness, but I have to at least knock off a possibility.
-
I think I misunderstood… Syringe to syringe with no storage medium to build/ store the pressure over time will not make it last longer. The energy in the wind if stored can be released quicker to do more work but for a much shorter time. An example of this is a typical phnumatic system, but there is no energy advantage, you don't get out more energy than you put in… ignor me and carry on lol.
ED. Please don't try to pressurise anything to 3,000,000 psi... It won't end well.
:)
-
Lol, robots running on 3000000 psi would be like robots running on nitroglycerin. Maybe a closed system with a liquifiable gas would offer similar storage capacity at lower pressures. Or just go with batteries. But then you'd need to increase the voltage at the cost of current to maintain constant charging. That might include DC to AC conversion then a transformer, then AC to DC conversion to charge the battery. The transformer would have to be adjustable somehow.
-
Yes, petrol is amazing that its a liquid, you get more for you buck that way. I bet if they ever made an efficient petrol -> electrical exchange it would be a bit like nuclear energy already. :)
So, I may be wrong - but there still may be a way to get to 3 million psi with your hand, but it has to involve shifting very little volume?
So instead of kheepings growing fatter indefinitely, you actually go big syringe small syringe big syringe small syringe, this will reduce volume as it increase torque, we sacrifice another dimension for the exchange, then it has a chance of happening again, albiet with more modest results.
-
Yup, you can get "any" pressure by hand, depends on the piston diameter. Highest pressure would be by pressing atoms together in single file .......................... like these dots. Might even achieve fusion pressures that way, don't know. Basically, Work (piston 1) = Work (piston 2) ---> Force*Distance (piston 1) = Force*Distance (piston 2). So you really only need one piston and then a holding tank. Extra pistons in series would just cause friction losses (I think).
-
So- Hopefully Something, u actually sound like you maybe understood things in highschool physics class.... - can I get a professional solicitation about this schemmy here->
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/70503266_998497733831617_6775140661231353856_o.jpg?_nc_cat=103&_nc_oc=AQn4TQsyKfLgC7jxTQy320OMVNndY3mM1WE7Kvm6cvbjK66_ZcgY1LmZ8U-UIVYQFBw&_nc_ht=scontent.fper6-1.fna&oh=81d961875171a0bbf674bbc531d7dae8&oe=5DF53E21)
So if I were to start from a very weak pump - to go into this thing, it might not be able to get over the force required to move the separators/weights.
But ive got an idea, if I make a tree - I can make 4 light pistons hit 1 heavy piston, it will be able to take the resistance better. So that would be losing the volume twice, or... do I not lose the volume the second time?
[EDIT] No. I think you have to lose more volume to take the resistance of the separators [EDIT]
-
Yes, they are the same from a computer simulation standpoint of the principal physics. Same force requirements and resulting volume loss. Physically though, X2X2X2 would be larger, heavier, require more moving parts, and lose more energy to friction. For this trouble it would give you higher efficiency at high flow-rates, because of the opportunity to cool compressed air at each stage, thus reducing the force required for the subsequent compression. I think X8 is the optimal solution for trickle charging from constant low intensity natural resources like the sun and wind, as we are planning. The primary piston of X8 could be ~1/5 it's length (same as X2X2X2 first piston) and still work more efficiently than the whole X2X2X2 contraption, while being lighter, and less prone to failure. That's a clear win for X8 if you ask me! (Unless you are dealing with crazy high flow-rates, or insulated temp sensitive air tanks.)
-
hmm good thanks for that, very impressed! ::)
I think the single cylinder method definitely sounds a lot simpler to make.
But im a little confused -
If u were going micro a 10 micrometre radius pump would be x1000 psi? (P=F/A?) ????
The multi-syringe model would have to have a needle at the end into the battery though if you put enough in a row, because the stroke length would reduce to nothing and the diode wouldn't open, because of air sponginess... but ive got my pneumatic conduction idea to help me transfer the power properly without fluid. (pump its atmosphere up.)
Its such a tiny little spek of a diode!! it goes into with near 0 play, to the battery, but it also has to support the backpressure, so maybe ud have to use something special like titanium film.
-
But im a little confused -
If u were going micro a 10 micrometre radius pump would be x1000 psi? (P=F/A?) ????
Area = pi*r^2 = 3.1415926536*(10/1000,000 m)^2 = 3.1415927e-10 m^2 = 4.869478424e-7 in^2
Pressure = (Force/Area) = (1 pound force / 4.869478424e-7 in^2) = 2,053,608 psi
:o
-
I couldn't follow the budget so well, but 2 million pounds of force! that's an optimistic one.
So I take it you can pop a car tyre with a skinny bike pump?? dunno...
I drew up a simple diagram of this pump im going to make to try and do this - the plastic or whatever needs flexation for the inlet and outlet, the outlet is flush against the piston, if this isn't the case then it wont hit full power against the opening to sneak the crack of high pressure air in.
(second drawing - got it flush in and out.)
(https://scontent.fper6-1.fna.fbcdn.net/v/t1.0-9/70933391_998652900482767_1899492867689152512_o.jpg?_nc_cat=109&_nc_oc=AQkvDi2avXj0CBXi2VsgzIpihXIyXKDaO0H9sMaYTFKFZ8wsRDL9jiqWL9_SGnv9KXY&_nc_ht=scontent.fper6-1.fna&oh=dc8508d81066f5dada3c769cd9c4e169&oe=5E05528C)
-
If you have the right pump you sure can pop a car tier with your arm strength.
-
Some Japanese researchers have done a really good human-style leg. Kengoro and Kenshiro are full-body musculo-skeleton robots that exercise.
This video shows the leg part at 0:20 sec. https://www.youtube.com/watch?v=6JNq1COqB_s (https://www.youtube.com/watch?v=6JNq1COqB_s)