My Robot Store

My Robot Store
This is my robot store! Click on picture above for more information

Friday, October 29, 2010

My theory on why the Bioloid GP has an unstable running gait.

I wish that I could get my hands on a GP for a few weeks. I think that I know why it is so unstable while running. The robot is under striding. The speed of a runner is based on their stride length divided by their turnover rate (the number of times your feet strike the ground per minute.). A runner needs to create a balance between stride length and turnover. You don’t want to under or over stride and you don’t want to have too high or too low of a turnover rate for your stride. The Bioloid GPs in the videos below use a shuffle step which means that their stride length is very small. The AX-18F servos give it a very high turnover rate which causes the unstable run. So the solutions are to slow it down a little or to increase the stride length.  It looks like the Bioloid Premium kit robot is faster than the GP in the videos below even thou it only has AX-12+. I think that this is because it has a longer stride length than the GP. So my solution to the slower speed and unstable run of the GP is to create more of a balance between the turnover rate and stride length. It looks like to me that roboticist have not yet adjusted to the faster speed of Dynamixel servos yet, specially the AX-18Fs. The shuffle step is great for slower servos but with faster servos it slows the robot down and makes it more unstable.

This is also a good tip to remember "don't lean forward -- that will cut your turnover rate by causing your leg muscles to tighten. Stay upright, with your head, shoulders and body aligned." This may not be as important for robots but good balance will increase speed of the gait.

This is a perfect example of what I mean!

I think that the TinyWave has the same problem too!

I think that Asimo has a very good running gait, well balanced with a good stride length and turnover rate.


  1. The basic problem is that they aren't running, and don't walk. They just shuffle their feet quickly, making tiny movements while maintaining balance (most of the time).

    You're correct about the stride/turn-over ratio. It's totally out of wack for a reasonable run motion. But, I really wonder if the design is capable of maintaining sufficient dynamic balance once the stride/turn-over ratio is closer to a human run cycle.

    ASIMO, as you pointed out, is much better optimized, but still cheats a bit.

  2. All that I know is that it would be fun to find out!