Modeling and attitude control strategy for a parameter-uncertain wheel-legged vehicle with a new reconfigurable suspension

Abstract

This paper focuses on proposing an active rolling mechanism and investigating its attitude control performance in wheel-legged vehicle. Unlike conventional force-controlled active suspensions, this mechanism forms a reconfigurable suspension in series with spring damping. This reconfigurable suspension can filter out high-amplitude vibration and allow for significant attitude adjustment. The kinematics and dynamics of the active roll mechanism are analyzed and a 16 DOF nonlinear dynamic model are presented for the whole vehicle. Due to the incomplete prior information about the lateral acceleration of the vehicle, a novel active roll strategy considering zero-moment-point(ZARC) is used to calculate the expected attitude motion depending on different scenarios for improving ride comfort and handling. Considering parameter-uncertainty problems, a reinforcement learning controller is designed to track expected roll angle in real time. The investigation of series simulations and tests with the proposed controllers will show that even if the parameters in active roll vehicle have variation, the performance of roll motion tracking control still can be improved effectively with the enhancement of lateral sensitivity and safety.