High-Speed Cornering for Autonomous Off-Road Rally Racing

Abstract

High-speed cornering can typically be performed only by an expert driver. Advanced control system design for (semi-) autonomous vehicles requires an understanding of such aggressive driving maneuvers in order to be able to take advantage of the full handling capacity of the vehicle, resulting in enhanced stability and improved safety vehicle characteristics. In this article, we first learn a primitive high-speed cornering maneuver using a series of demonstrations obtained by solving the minimum-time cornering problem subject to different initial conditions. This primitive trajectory indicates that a typical high-speed cornering maneuver can be approximated by three segments, namely, entry corner guiding, steady-state sliding, and exiting. Based on this observation, we divide a high-speed cornering maneuver into three stages. A switch-mode control strategy is designed for these three stages, using a combination of linear and nonlinear control techniques. Conditions for fast trajectory replanning, optimal speed profile generation, and tracking control are analyzed. Simulation and experimental results validate the proposed approach on a fifth-scale robotic vehicle.