FPGA-based Controller for Kinematic Systems

[wuselfuzz]

Short IK demo video:

http://fnord.terror.de/ik_demo.avi

Powers up with a predefined position with legs in the air, moves up, moves legs to idle position, rotates body around
x,y,z axis, after that around a few axes at once, then moves back to idle position, sits down and powers off servos.

Two walking algorithms are also in the works, but I forgot them on my other computer.

[wuselfuzz]

Added a battery pack from four spare 1700mAh batteries today. Charged them at 500mA for 2.5h and connected them to the servos. With a demo where the bot lifts legs 0,5 and 1,4 and 2,3 alternating, it ran for about 20 minutes which fits my expectations.

With another 4 cells of those batteries on top, total weight is 1960 grams.

[wuselfuzz]

Update:

I'm in Boston currently, visiting my gf who works as a research fellow at Children's Hospital until September and I brought my hexapod in the luggage. My luggage got lost and I got it back on the third day after being here, the hexapod is in order, but somehow they managed to pull out a couple of atmega162s from the antistatic foam they were in and bent all the pins. At least they're not expensive.

On the technical side, I eliminated another stupid bug from my IK routine (I was rotating the leg base offsets instead of the end effector positions) and my walk() function now allows to walk turns.

I also pondered about pressure sensors in the feet and had an idea for a very low-cost solution: I'll try putting carbon powder into those rubber feet and measure the resistance. This should work like a microphone, when pressure on the powder changes, the resistance should change.

[Matt Denton]

Hi,

Well it was a good thing you got your hexapod back!

I also pondered about pressure sensors in the feet and had an idea for a very low-cost solution: I'll try putting carbon powder into those rubber feet and measure the resistance. This should work like a microphone, when pressure on the powder changes, the resistance should change.

This is a similar idea that I have been pondering.. I would be interested to see how this works out. So do you plan to cast your own rubber feet impregnated with powder?

[wuselfuzz]

This is a similar idea that I have been pondering.. I would be interested to see how this works out. So do you plan to cast your own rubber feet impregnated with powder?

I just finished a quick proof-of-concept design literally made from office supplies.

I killed a pencil and ground the carbon lead for carbon powder.

Then I got a second drinking straw at lunch that fits exactly into the rubber feet you provided with the MSR-01 kit. I cut off a part of about 15mm.

I also cut off a piece of a paperclip and bent it to a J, attached it to the outside of the straw with some invisible tape and also closed the lower part of the piece of straw with that tape.

This was put into the rubber foot. I put a second L-shaped piece of paperclip into the straw and soldered two wires to both pieces that run to a multimeter.

The whole construction looks like this (fear my ascii art skills):

Code: Select all


   foot of hexapod goes in here
     |
     v
   \      /
   !|   !|
   !|_  !|
   !####!|
   !####!|
   !####!|
   !###\_|
   =======

The whole construction goes into the rubber feet supplied by the MSR-01 kit.

!   = drinking straw
#   = carbon powder
\_| = bent paperclip
=   = invisible tape


For a start, the resistance is larger than 10 ohms (about 15) when the foot is in the air and less than 10 ohms (about 7.5) when the robot is standing. Unfortunately, I'm a bit challenged electronics-wise, because I didn't bring very much stuff on my vacation. But at least I got some results for this idea already.

[Matt Denton]

Hmmm... well I'm more than impressed with what you have achieved with a pencil, a paper clip and a straw!! but I'm still a little confused of the construction, any chance of a photo?

[wuselfuzz]

Hmmm... well I'm more than impressed with what you have achieved with a pencil, a paper clip and a straw!! but I'm still a little confused of the construction, any chance of a photo?

I can take a photo tomorrow, left the bot at my gf's workplace. But I'll take it home on the weekend or I could also take the camera there tomorrow.

Now I'm pondering about how to actually measure that resistance without having a large current running through it, since the resistance is pretty low.

Another low-tech idea would be diluting the carbon powder with fine grained sand, an electronics solution would be using an opamp obviously. But unfortunately, I studied CS and not EE, so I don't have much of a clue about more complex analog electronics.

However, a friend suggested sampling some interesting ICs from TI.

I already fitted all 6 legs with this ghetto-type sensors, all are less than 10 ohms with pressure applied and about 50-100 ohms without.

[wuselfuzz]

Images:





In the second image, the carbon powder is missing, because I didn't feel like killing another pencil just for pictures and it's hard to get them out of the rubber feet again without destroying them. So I just made another one for the pictures.

[Matt Denton]

Ok, its starting to make sense, It looks as though the two paper-clip parts could short across the aluminium tibia, but I guess there is some insulation between them. I'm still very impressed with what you have managed to achieve with a paper-clip, straw and a pencil..

[wuselfuzz]

Ok, its starting to make sense, It looks as though the two paper-clip parts could short across the aluminium tibia, but I guess there is some insulation between them. I'm still very impressed with what you have managed to achieve with a paper-clip, straw and a pencil..

Well, one of the paper clip parts is completely inside the straw. I also probed if the aluminium is conductive, but it seems that the eloxal layer made them insulating.