PDA

View Full Version : 2 DoF Quad gait



ArduTank
07-31-2013, 09:39 PM
Okay guys, all I'm asking for is information on the best ways to create a gait for a 2 DoF Issy Dunn Yet style mech. I will be using AX-12A's for servos.

I have asked Lnxfergy, but have gotten no reply.

jwatte
07-31-2013, 11:11 PM
I only know of one way. The "ankle" lifts up/down, and the "hip" does locomotion. This leads to foot sliding.

Start with FL and RR hips turned "backwards" and FR and RL hips turned "forwards" and all feet down.

Now, FL and RR ankles turn "up"
Then, the FL and RR hips servos turn forward, and the FR and RL servos turn backward. This uses the FR/RL pair to push the bot forward.
Now, FL and RR ankles turn "down" and FR and RL ankles turn "up"
Then, FL and RR hip servos turn backward, and FR and RL turn forward. This uses FL and RR pair to push the bot forward.
Now, repeat from top.

Turning "up" should probably be less than an inch upwards, because you will likely "skate" on one of those two "up" legs, and you want to not tip too much backwards/forwards during this movement.

It may be possible to twist the ankles that are in contact with the ground to stay at a relative distance sideways from the bot center, and instead move the body up/down, to avoid foot sliding (instead getting body up/down motion) but I haven't worked out the math for that or know anything about whether that will actually be an improvement in capability.

ArduTank
07-31-2013, 11:29 PM
Yeah, I used that technique with M-O in that junk walking video. I finally realized that if you added surface area to the foot (make it a square or larger circle, instead of a little peg) that the bot might walk without foot sliding.

jwatte
08-01-2013, 12:41 AM
the bot might walk without foot sliding

The only way that can mathematically work is if you also alter the contact angle with the ground at the same time, and thus raise/lower the body at the same time. And with a "spike" foot, that should be equally possible -- as long as you can get sufficient friction between foot and ground.

The reason I say this is because, if you don't tilt the foot while sweeping the hip front-to-back, then the arc of the foot contact point around the hip joint is a pure circle. And the relative X position of the foot will then be leg-length times sin(angle) (where angle is 0 when straight forward.) This is what forces (sideways) sliding.

So, if a 2dof foot can reach forward to point X with "full extent," and you can then compensate the radius of the leg while it's sweeping, by tilting the foot down (and thus moving the body up,) then you should be able to avoid sideways foot sliding. I have not tried it, though.

ArduTank
08-01-2013, 07:38 AM
Oh, okay. I didn't know what you were meaning by foot sliding. I meant the bot tilting forward and the foot that is lifted on the front ending up being a skid plate.