I fancied making some robot legs but don’t really have access to equipment or facilities to do this but I stumbled across a program called linkage2 on the Internet and decided to have a go at designing some.
Just as well I didn’t start off with some sketches and a few hours in the workshop. As it turns out getting all the right dimensions to make the legs do exactly what you want is very, very difficult. I stopped after the 16th significant design change of which there were hundreds of small incremental changes along the way.
If, and it’s a big IF, this was to go any further it would be time to start doing some real world tests...
These were my objectives:-
- The legs would be powered by a rotating motor and not servos or actuating cylinders. The reasoning here was that servos and actuating cylinders have to stop and start in the opposite direction simply to make the legs keep moving forward. This is more mass going through changes of direction and I hoped that a steady rotating driving force would be able to have less mass changes even with control rods.
- The legs would travel on the ground in a flat horizontal plane so the robot body would not rise and fall as it walked/run.
- The down time of a leg was to be 50% of the walk cycle time so there was no scuffing of the second leg or no time when both legs were (partially) lifted. Because the leg not supporting the robot has to rise to clear the ground it has to travel further and faster on the return to position itself for the next step.
The gait could be changed from standing to walking to running by changing the control rod lengths and/or positions. The driving motor could be kept turning to allow robot speed to be determined by gait/step size and also by the speed of the motor. You can’t see it in the videos but there are controls on the model that allow these things to be changed as it is goes through its cycle.
I was pretty pleased with the outcome and I show in the video links two of the solutions in the next step. The first one shows two legs which would normally be side by side but I have moved one forward to make it easier to see how they work together. In other designs I tried put a ‘foot’ on the end to articulate in a way to assist in shock absorption as the foot falls and assist in springing away.
The second video shows another design with the legs side by side and you can see there is no time when the legs are travelling horizontally so the robot would bounce as it ran. This may be a good thing in practice and that is where some real world testing comes in.
A long way to go before DARPA or Asimo have to worry but a nice intellectual exercise all the same.