Path tracking control of a self‐driving wheel excavator via an enhanced data‐driven model‐free adaptive control approach

Abstract

In this work, an enhanced model-free adaptive control algorithm considering the time delay (EMFAC-TD) is proposed for a class of multi-input and multi-output systems and is further applied to the path tracking control problem of a self-driving wheel excavator. First, a preview-deviation-yaw-based method is first proposed for the simplification of excavator dynamics. Then, based on the simplified dynamics, the EMFAC-TD controller is designed via a novel dynamical linearisation technique with a time-varying parameter termed pseudo Jacobian matrix, which contains the coupling information of the excavator driving dynamics. Moreover, for the time delay of the excavator, EMFAC-TD method also handles it by introducing the Smith estimating method. The main feature of the proposed EMFAC-TD is that the controller is designed only based on the input and output data of the excavator. Further, the stability of the proposed EMFAC-TD is proved via rigorous mathematical analysis. The experimental results implemented on a real self-driving excavator show that the maximum tracking error of the excavator controlled via EMFAC-TD method can be limited as 0.3–0.7 m for different driving tasks, satisfying the practical requirement. The experimental results verify that EMFAC-TD method can improve the tracking performance and complete the required task successfully.