Model Reference Adaptive Control for Actuators of a Biped Robot Locomotion

Abstract

Many robotics problems do not take dynamics of actuators into account in the formulation of the co ntrol solution. The fallacy is in assuming that desired f orces/torques can be instantaneously and accurately generated. In practice actuator dynamics may be unknown and can have significant transient effect on the overall results. This paper presents a Model Reference Adaptive Controller for the actuators of a robotic biped walker. The actuators learn and produce the torque required for a walking cycle presented in the form of a torque reference model. The actuators of the biped adapt themselves to match with the desired torque reference model with the aid of Lyapunov stability criterion. This control scheme for the biped robot is simulated on a sagitt al plane. The results verify the MRAC scheme for the actuators.