Robust Path Tracking by a Dubins Ground Vehicle

摘要

The robust linear tracking technique is applied to a practical path tracking problem for a ground vehicle modeled as a Dubins car. The tracking control is designed based on the optimal strategy in an auxiliary linear-quadratic differential game for a linearized vehicle model. In contrast with a purified theoretic setup, the real-life problem is complicated by a nonconstant speed and control saturation. Moreover, the actual path generating control can be unknown to the tracking strategy designer. These issues required additional design blocks: speed tracking loop and leading angle differentiation, both based on similar tracking algorithms. The saturation effect is considered by using command reinforcement. Numerical and experimental results are presented and compared. Robustness with respect to unknown disturbances is demonstrated. Novel analytic results on the tracking error/control effort tradeoff are also presented.