Path Following Control of Tracked Mobile Robot Based on Dual Heuristic Programming

Abstract

Path following has been an important research topic in the field of mobile robot. The traditional path following algorithms need to tune the parameter finely by expert. It is difficult to design an algorithm which can be self-optimized with respect to different radius of road. In this paper, a path following method based on dual heuristic programming (DHP) is proposed. Firstly, mathematical model of tracked robot is given, beside that tracking error model based on preview-following driver model is also given. Secondly, a path following controller is designed based on DHP algorithm. The actor and critic module in DHP frame are implemented by neural network. Thirdly, two different reference trajectories: combination line of straight and curve line and helical line, are provided to verify the performance of DHP method. As comparison, the LQR algorithm is also used. The simulation results show that, on the combined line, the absolute value of mean error and root mean square error value of the LQR method are 0.04m and 0.27m respectively. While the absolute value of mean error and root mean square error value of the DHP method are 0.027m and 0.19m respectively. On the helical line, the absolute value of mean error and root mean square error value of the LQR method are 0.38m and 0.31m respectively. While the absolute value of mean error and root mean square error value of the DHP method are 0.16m and 0.27m respectively. These results show that the proposed method has better control performance. During the path following, the DHP method keeps on correcting its parameter to optimize control performance, which further indicated that the proposed method has more adaptive to different road conditions.