Sensitivity of Legged Balance Control to Uncertainties and Sampling\n Period

Abstract

We propose to quantify the effect of sensor and actuator uncertainties on the\ncontrol of the center of mass and center of pressure in legged robots, since\nthis is central for maintaining their balance with a limited support polygon.\nOur approach is based on robust control theory, considering uncertainties that\ncan take any value between specified bounds. This provides a principled\napproach to deciding optimal feedback gains. Surprisingly, our main observation\nis that the sampling period can be as long as 200 ms with literally no impact\non maximum tracking error and, as a result, on the guarantee that balance can\nbe maintained safely. Our findings are validated in simulations and experiments\nwith the torque-controlled humanoid robot Toro developed at DLR. The proposed\nmathematical derivations and results apply nevertheless equally to biped and\nquadruped robots.\n