I saw this little 3 servo hexapod project on Pololu and thought I would try to make one with a large platform and much more battery power.
[url]Pololu Micro Maestro Hexapod Robot - YouTube
First I made a virtual sketch to calc the weight and balance and see if the little SG-90 servo would be able to handle that weight. The center legs in that config will start to get strained at 180g so the total balanced weight had work out to around 300g.
[url]http://dl.dropbox.com/u/3647376/walker1concept.jpg[/url]
This is a fully loaded test with 70g on the platform but it can handle 100g just as well. The walking is more deliberate because it has to shift so much weight around.
[url]- YouTube
It’s headless and walking backwards for the test because it’s using an Atmel 328 and I don’t feel like programming more than a few lines with that. I just use those for tests because I don’t care so much about frying a $5 chip. Now I can start to piece together the final Fez version.
That is awesome! I like the intricate wire framing. One thing it made me wonder about is if having the batteries in the legs actually makes it harder on the servos as they are having to lift a weight at a distance rather than just shift a centrally located weight. One thing it did do was make the bot walk more naturally, the Pololu one looked a bit like a water bug.
@ Jeff The main leg servos only really take stress if it runs too fast because it would have to hold back the weight when it shifts. If it’s kept a smooth motion then it never really has to do any work. The center legs are able to stay shorter with the main wait to the outside so it doesn’t take as much power. I didn’t brace the center legs because they make a natural shock absorber and will fail before the servo is overpowered.
Are you going to use the batteries in the legs to power it?
Which FEZ are you planning on using? Mini?
Are you going to use the IR sensors like the Pololu video?
I finally plugged the mini into it and programmed the 4 main movements. It’s simple FOR loops right now so the different states aren’t aware of previous positions yet and it jumps into place. There actually has to be 6 movements total because left/right rotation would each have CW and CCW versions.
[url]- YouTube
It really just needs one IR sensor on the front because it sweeps 60 degrees every loop. It’s also not as fidgety as the Pololu version and can be more confident of its surroundings and heading.
I added the main sensors so it can navigate by itself now.
[url]- YouTube
The front IR sensor is angled down slightly and can double as edge detection if the edge is within 45 degree of the center. 2 photo cells let it know to rotate towards light and help prevent it from getting boxed in a corner.
Finally have it assembled and running under its own power now. It’s 50g lighter without the breadboard and much more stable now.
[url]walker1assembledTest.mp4 - YouTube
It’s using 2AH hybrid NiMH batteries for power so it should be able to run at least 5-6hrs.
![http://dl.dropbox.com/u/3647376/walker1concept.jpg[/url]](http://dl.dropbox.com/u/3647376/walker1concept.jpg%5B/url%5D){kind=link}
![http://dl.dropbox.com/u/3647376/walkerLeg.JPG[/url]](http://dl.dropbox.com/u/3647376/walkerLeg.JPG%5B/url%5D){kind=link}
![http://dl.dropbox.com/u/3647376/walkerPlatform.JPG[/url]](http://dl.dropbox.com/u/3647376/walkerPlatform.JPG%5B/url%5D){kind=link}