Implementation of the SIMBICON paper.

Here's a video of a running biped.

Simbicon Controller

In order to visualize a motion, specify the target angles and other parameters in a .yml file. (See settings/config.yml for an example of how to specify this file.)

Then run python simbicon.py with relevant parameters as specified in the simbicon.py file.

Example: python simbicon.py -m jog -p settings/config.yml

CMA Optimization

CMA is used to optimize simbicon parameters (target angles, torso_kp, torso_kd, FSM time interval) for a given target velocity and style. Look at cma.py for details on what parameters to specify before optimization. Running this file will yield a yml file with saved parameters.

To visulize the optimized parameters, you can run the simbicon.py file with the path to the optimized parameters.

Example: Here we use the initial parameters for jogging (which is at 1.8m/s) to optimize for faster running. python cma.py -lm jog -sm running -sp settings/cma_config.yml -tv 3.5 We can then visualize the result: python simbicon.py -m running -f cma_config.yml

Acknowledgements

Thanks to Ben Ling for the CMA optimization code and Michiel Van de Panne for helpful discussions.