Hexapod Walker

Paul Sherman

Hello-
I'm working on an R/C 6-legged hydraulic walker and I
wanted some input for using a Basic Stamp for controlling
the hydraulic spools with servos. I chose hydraulics for the potential for
integrating other systems, the look, the engine sound, general vibe.
Pics of the project thus far here:

http://personalpages.tds.net/~ksteuer/machine/

Given that hexapods walk by "alternating tripods", I'd like to skid steer the machine, but I need a way to get both sets of legs back in sync once the turn is done.

So it would be two valves (under servo control) acting as transmissions (forward, neutral, reverse) controlling two hydraulic motors which control the legs via drive chain. I can have either each side controlled by one motor, or I can have the tripods (2 legs on one side, one on the other) driven by the same chain.

I was going to get the machine back in sync just by nudging the servos from a transmitter, but could I use a microcontroller to keep everyting sync'ed together?

Any help or direction here is appreciated.
Thanks
Paul

Post Edited (Paul Sherman) : 11/28/2006 6:13:10 PM GMT

LSB

Paul,
I think what you propose is possible; certainly hobby type R/C receivers and stamps are compatable. The output pulses of the receiver are fed into the stamp through a 220 Ohm resister as I recall and the stamp responds to the pulse width as it would any other variable. My experience is limited--I once wired up one R/C channel to activate five LEDs-- input of 1 to 1.2 mS pulses turned on #1, input of 1.2 to 1.4 turned on #2, etc... You might look at servo output boards if you intend to do a lot of interface--the stamp isn't very fast (relatively) and servo output of several signals tends to bog-down the processor. My project never went beyond breadboarding, but I've seen several sites where R/Cer's have gone much farther with this concept. I imagine that with some work, one could build a very functionable 'fly by wire' system where the stamp interpreted ALL the receiver signals and completely managed leg (or pincer, death ray, ...) control-- even playing pre-recorded macros or responding automatically to sensor input.

Cool project by the way-- at a PARTS Robothon I saw one of similar size, but more crab like. As it activated and the pnumatics lifted the body to knee height people were wide eyed with awe---me? I ran, I know how tricky these things are to control!

Paul Sherman

Thanks for the reply-

I realize there is all sorts of control potential here, but what I want now is real simple until I become more adept.

As of now, there will be only TWO components to control, the hydraulic valves (see pic). One per side, controlling each side. Forward, neutral, reverse, via some big 1/4 or 1/2 scale servos. That's it.

All I need the microcontroller to do is maintain sync between the two sets of tripods. So if I wan to skid steer it like a tank, that's going to throw the cycles off; I want the controller to essentially tell one side to "wait" until the other side is in sync, in other words (see other pic) if Right Side B legs are down, they don't move until Left Side B is also down (I realize position sensors will be involved...)

Speed sync might also be nice; perhaps an RPM sensor so keep each side within reasonable range of each other...

Thanks
Paul

LSB

Yep, possible. Perhaps a position sensor (switch, optical device, or potentiometer) to measure leg position (one each R/L channel should be sufficient) as stamp 'watches' incoming pulses. while "turn" 'stick is centered at 1500 mS BS can interrupt/slow servo pulses to sync the legs. When turn 'stick varies from center, stamp allows out of sync turn, until 'stick returns to center position, then syncs legs again. I see this as being useful to in eliminating inevitable mechanical drift during forward motion as well. Kind of an 'If R leg channel is in position when L leg channel is position then SYNC=true else Stop (or slow) Left until R leg channel sensor indicates correct position to start left channel' type of arrangement. More complicated schemes may slow the <I>leading</I> channel or move "UP" triad of legs to the 'start step' position before releasing control.

Re-read your post. I think the point is that the BS can easily intercept, interpret and output servo control pulses. How you measure leg position and what pulse chains (servo positions) the 'Stamp supplies are up to your imagination. I only caution that many servos or complex, highly accurate movements may bog down the processor with timing reguirements; in that case a serial servo control board will handle the timing for you.

Post Edited (LSB) : 11/30/2006 2:31:59 PM GMT

Paul Sherman

Beautiful. Exactly what I wanted to find out. Thanks for the input.

Looks like it's time to stop welding and learn some PBASIC.

Thanks again
Paul