During autonomous car racing, the car should accurately, quickly, and safely navigate the path while recognizing other cars and objects on the road to avoid collisions. This study aims to design an adaptive path-tracking controller that can avoid collisions in real-time while driving across the entire course in a robot racing competition. An embedded PC with an RTK-GNSS and 3D LiDAR is installed on an ERP42 racing robot platform, which is used to test the autonomous driving system. The designed system, consisting of a GNSS-based autonomous navigation system and a LiDAR-based collision avoidance system, utilizes four functions for its operation: curvature-based speed control, variable parameter tuning, deceleration, and lane change. Each function is tested at an experimental site to evaluate its performance and functionality. The RMSEs of lateral deviation and heading error are obtained by comparing the trajectories of the robot in a given path. The standard deviation of the steering angles is also calculated to evaluate the stability performance of the robot in the field. LiDAR is found to be effective in avoiding collisions with other cars and objects installed on the road and facilitates changing the traveling lane while effectively reducing the velocity.
Use of LiDAR and GNSS for Collision Avoidance-Based Adaptive Path Tracking in a Racing Robot
Lect. Notes in Networks, Syst.
International Conference on Intelligent Autonomous Systems ; 2023 ; Suwon, Korea (Republic of) July 04, 2023 - July 07, 2023
2024-04-25
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Optimization-based path planning and collision avoidance for autonomous racing
DOAJ | 2023
|