Online regulation of the walking speed of a planar limit cycle walker via model predictive control

Limit cycle walkers (LCWs) are biped robots that exhibit a stable cyclic gait without requiring local controllability at all times during gait. Online regulation of the walking speed in LCWs is challenging because their walking speeds cannot be directly controlled or preplanned as in the case of fully actuated robots. In this paper, we propose a novel idea of using model predictive control to achieve online speed regulation in a planar LCW. The system we designed has a two-level structure. The walking controller at the low level is composed of state machines. It only accounts for the motor control of the actuated joints in the robot. Nevertheless, the model predictive controller at the high level regulates the robot's walking speed online by modulating the parameters of the walking controller. It is demonstrated in real-time experiments that the walking speed of our robot can be regulated online with this control structure.