Trajectory Optimization and Robust Tracking Control for Off-Road Autonomous Vehicle

Abstract

This paper presents a control strategy for real-time trajectory optimization and robust path tracking for unmanned off-road vehicles to ensure both stability and performance. The approach takes advantage of a two-degree-of-freedom control framework that combines predictive driving control through perceptual information and feedback control for robust stability. Trajectory generation leverages model predictive control where the particle swarm optimization is used as an optimizer to address problems of the non-smoothness of the traversability information and nonlinear nature of vehicle dynamics. By using the exteroceptive perception, the vehicle could estimate traversability and adapt its motion to achieve fast and smooth driving. For the feedback controller, a system level synthesis is used to faithfully track the planned path despite uncertainty and unknown disturbance. Specifically, we focus on realizing the proposed method as a practical means for the real-time control system. The effectiveness of this method is validated through extensive numerical simulations and experimental tests, demonstrating its practical applicability in uncertain environments for autonomous vehicle navigation.

Keywords

• Autonomous vehicle
• linear quadratic regulator
• model predictive control
• vehicle dynamics
• convex optimization
• robust control