Virtual Passive Bipedal Walking Based on Model Reference Adaptive Control

Abstract

Passive dynamic walking has the advantages of low energy consumption and natural gait, but it can only be realized on a downward slope with small inclination angle and is very sensitive to external disturbance. Therefore, we combined passive walking with adaptive control, and designed a virtual passive walking control algorithm for biped robots based on model reference adaptive control. In this control algorithm, the reference model is a virtual robot that walks fully passively in an ideal condition, and the gait of the actual biped robot tracks the fully passive walking gait output by the reference model. The adaptive control law is derived by Lyapunov methods. The simulation results show that the proposed virtual passive walking control algorithm can make the biped robot walk stably on the upward slope or irregular ground and has certain anti-interference ability.