[goaty]

So this is what goes in the centre of the robot, and its leg come out the edges of the image, and that's half the actuator done, the other half is an amplifier and then the main pump, which has a bigger pipe than the logic does,  but im needing a map way bigger than what I thought, its about 2500x2500 -  this is a very clean machinable image,  its made with "Mario" tiles,  and its going to have air running through it.  Ive got it all down in theory, logic gate is organized, I just have to build the map now,  theres a tiny bit more to go on the editor, but we cant be bothered so we are doing a bit of designing instead of doing what im supposed to be doing.



Ill post more here as I build it.

(addendum)
Pneumatic Motor driver v0.12, finished, except for some little via bugs here and there I haven't spotted yet.

[goaty]

And! howabout this! 


And now I added the amplifier to it,  so you can see theres something inefficient about the way im doing things...    But that's nearly it -  but ive got a problem,  this next "wire" im adding to it is the main thick one that powers the leg,  so I need to do a bit of theory first.


Heres an anatomy of my work.  (its the motor driver. but theres no motor, its a muscle driver organic.)


And if this is just the muscle drivers,  the brain is going to be a real headache...

[goaty]

Here it is done...  but this is looking WAY too detailed!!  Ill never be able to fab this...    I was hoping to get a 2.5cm spider out of it,  but this is way too much...  this is a 5000x5000 map, and I cant build it like that,  even professional 3d printers don't even pack that much in a single layer image...    so...   I think theres something wrong with my theory.



upclose of the finished one...  im scanning it now for bugs.    but the whole thing could possibly be wrong, as ive not done enough tests...





If I put the motor drivers for each leg ontop of each other makes about 2 thou x 2 thou,  which is more doable, but I think this thing has to caste by itself,  and I have to make the robot bigger,  but its very flat... this is a flat circuit.      And I think I might have problems with the air making it through the pipes, too skinny.

[goaty]

This next pic might make things a bit clearer what im doing...

[HS]

If its flat already and pneumatic then you can stack lots of these to get more air flow right? Or have it control valves from an air tank instead of muscles directly. If you find a way to make one good casting mold (it'll have to be an inversion of this) then you can probably pour out lots of copies. This is amazing wizardry. How? What did you base the design idea on?

[LOCKSUIT]

Goaty, how about you run a simulation of it in a wind simulation in say Blender? Blender is EASY to learn. Yes Blender has some wind sim. All you need do is import your image and raise the black ink up to make walls, then put 2 plates to sandwich it. You can run multiple designs.

[goaty]

Its a 500x700 bitmap so the wire ends up too skinny so not much air flow,  but indeed it is flat so you could make it as tall as you want - But I think ive got the solution to making it simpler, so hopefully itll go back to 256x256.    It does look a bit too much for an analogue motor driver.

Indeed I want to casting mold it to get many copies     Its solid state,  except the membranes of the capacitors fill and empty of air pressure + the diodes are sucked hinges.  Its all held with pressure so even if I used syringe like widgets it will hold its state even if you shake it!  cool.

The logic itself is based upon resistors, capacitors + diodes, all pneumatic, no electricity,  its unusual, my original concoction, has only flex movement other than that its just solid state and very moldable, and it has a weird way of amplifying and this motor driver ended up getting too big.  If you look real close you can see little flats with barriers next to them, and little letters,   C- stands for capacitor,   S- stands for split into 2 halves Top and Bottom,  And T AND B is just the top guarded off, and bottom guarded off,  and I also use that to do the vias.

Be my guest if you want to go make your own logic paradigm for a machine!  Id be good to have around for advice,  why fork out for 100 arduinos and motor drivers when you can make them yourself.

I found 1 via layer is all you need,  it seemed to work nicely.


LOCKSUIT - That's a good idea,  yes simulation, but I think its something I'd have to code myself,   it would be hard to find a pneumatic sim that would suit my needs.

[HS]

I was looking at videos of these soft robots nasa is working on, and I was thinking a pneumatic computer chip would present a novel opportunity to incorporate an analogue response, or self correction. If this was cast in a flexible material like silicone, you could get some unintended but potentially useful consequences. Like if a place was too narrow, then it would expand, and maybe even optimize itself in unforeseeable ways, as analogue circuits have been known to do in the past. Effects can spread out and interact with each-other in surprising ways.

[goaty]

I was looking at videos of these soft robots nasa is working on, and I was thinking a pneumatic computer chip would present a novel opportunity to incorporate an analogue response, or self correction. If this was cast in a flexible material like silicone, you could get some unintended but potentially useful consequences. Like if a place was too narrow, then it would expand, and maybe even optimize itself in unforeseeable ways, as analogue circuits have been known to do in the past. Effects can spread out and interact with each-other in surprising ways.

Yes,  if it were brute forcing its movements, it could most definitely choose to morph itself if it suited its utility function.   This unit i've got planned for this project, has 2 morphs it can do,  it can extend its leg out like stretch-arm-strong, or contract it, sucking into itself,  its not quite useful and its just a side effect but it can do it,  and it will actually do it if it suits its motivation.

Just an update,  ive got a major rework to do which is going to reduce complexity a lot!   (I may be able to fit all 24 actuators on a 128x128 tilemap!!!)
Im doing it by sitting the tunnels right up against each other, and making it so I can via over diodes, resistors and next to the capacitors, and then it should be just the count of vertical lines going across and horizontal lines going down which will be the finished size I need.

So ill be working on that today,   I cant wait to see what it ends up looking like,  maybe a condensed maze- all via'ing ontop of itself.   but its still doing in just a high segment and a low segment, and no more height is taking up,  its actually still all flat, so theres plenty of room to tower for some interesting activity/brain activity.


last brainstorm->



Its going to look a little different,  I think ill use multicolour for this new one,  and itll look cuter.