Options for AX-12 Robot Legs

Posted by Robots Australia | 11:04 AM | | 1 comments »

While I wait for the Dynamixel Robot Controller PCB to arrive from the board manufacturers (update: they got here this morning! looks great!) I have returned to the mechanical design of the Quadruped4 robot.

I was never really happy with the look and function of the leg and hip design I have to date for the little creature, and thought I would create some alternate concept simplified cad models to look at other possibilities.

This is how I last left the mechanical design of the legs. The two axis shoulder or hip joint is implemented by two Dynamixel hinge brackets end to end bolted, at ninety degrees to each other. The thigh is made up of the two AX-12 servos end on to each other.

My main complaint about this configuration is the distance between the two axis making up the shoulder, which tends to make the upper part of the leg look out of proportion if it is fully extended downwards.

Alternate Concept 1
The first thought was to perhaps mount the forward swing servo onto the horn of the other hip servo, this allows really good thigh range of movement, but the bushes in the servo probably won't like all the load on the horn. Better to use the passive hinge point on the other side of the servo as well with one of hinge brackets.

Alternate Concept 2
Robotis manufacture an offset hinge bracket which I thought might be more useful than the straight bracket in allowing a greater range of motion of the joints given horizontal mounting of the thigh servo (as shown left). The image shows crouch and full reach (both dimmed) with standard stance pose undimmed.

Alternate Concept 3
I turned the offset hinge joint around on the thigh to hip joint to see if that improves the range of motion. I don't think it does (see the dimmed forward reach motion in the picture to the left) as the limb can't reach forward particularly far. The crouch pose does look a bit more 'natural' I suppose.

I can't say I am really satisfied with any of the options thus far; more cad experimentation required!

I have been grabbing the occasional hour here and there and have finally finished the layout and routing of the Dynamixel Robotics Controller PCB. A little bit of clean up and it will be ready for submission to the board manufacturer. I have to get going on ordering the parts for it while the board is getting made.

On the web site front, I have put together some info for the page about the Quadruped4 design which uses the Dynamixel Robotics Controller. Hopefully it can help other experimenters find suppliers or gather useful information from the approach I have used.