Hybrid trajectory planning and control architecture in high speed autonomous racing

Abstract

We address the path planning and control architecture of autonomous racing in high-speed environments. Path planning for high-speed autonomous racing differs significantly from approaches used in conventional urban environments. Unlike urban settings, there is no lane information, traffic rules, or signals, and there is no cooperation among drivers. In this thesis, the limitations of algorithms used in urban environments are recognized, and a novel approach is proposed. The path planning module incorporates interactions with surrounding opponents by utilizing their predicted trajectories. Previous research has been conducted under constrained conditions, such as head-to-head racing or rule-based racing on oval tracks. This study develops a path planning and control system architecture for road course racing where rules with opponents are not predefined. The proposed architecture’s performance is validated using simulation and real-world testing with the IONIQ5 production vehicle.