- Published: Monday, 17 August 2015 03:44
- Written by Kyle Granat
This isn't the First Time I've tried to get a RobotGeek Crawler going. I toyed with the idea of a MK-II style crawler, and later a MK-I style quad (because of the lack of 45 degree brackets for RobotGeek). At one point I passed it off to Wade who tried some head-crab like machinations
None of that really went anywhere though. But last week I was thinking about what I could do to make a simple 2-degrees of freedom(dof) quadruped. So that means each leg would get 2 servos, so a total of 8 servos. On your standard Arduino/ ATMEGA 328 you get 6 hardware PWM channels - so you can control up to 6 servos. There are lots of ways to do PWM in software like KurrtE's ServoEX Library, though it will take up some processing power. Still, being able to control 8 servos with no additional hardware is great for keeping the cost down.
I started off with a really simple design shown in the first picture. The body plates Kevin cut me for other experiments. Basically it's the body plates for the PhantomX Quad with the centimeter grid pattern and servo slots for RobotGeek Servos. The coxa to femur is just a dual hinge bracket. The femur extensions are actually plates from the RobotGeek Snapper Arm. Part of me wants to do some alternative designs (swap the coxa/femur ), but that makes mounting harder, so I'll probably stick with the current design.
It was the weekend, so rather than wait for Kevin to cut me some plates, I fired up the 3d Printer and made some plates for the femur extensions / feet. Just a simple taper, a little reminiscent of the MK-I feet. I'm still considering some other options (flipping the femur servo, making the extensions longer, etc) but it's a start.
So the million dollar questions is - Does it walk?. Well, no, not yet. I've got it moving around and doing some motions, but nothing too fun. I'm doing a lot of reading in Inverse Kinematics, and I've got a super simple solver for a 2dof leg to make it move to a position in xyz space. Now I'm trying to figure out exactly how I turn that into consistent movement on the crawler. I'm going to start with rotating the body and the move onto walking. I've been walking through the NUKE generated code that we use for the 3 dof crawlers, which has been useful - I just have a ways to go.
So if I'm still working on this next week, maybe I'll have it shambling around. I'm also thinking about a couple different walker type robots that might be fun, but I want to see where this one goes.