A Method of Parking Path Planning for Underground Autonomous Mining Trucks Based on Wd-RRT* Algorithm

Autonomous mining trucks provide an effective solution for safe production and efficient transportation in underground mining areas. The low-speed maneuvering process of underground mining trucks in loading zones can be regarded as parking behavior. Due to the characteristics of underground mining areas, such as confined spaces, steep slopes, and sharp turns, traditional ground-based parking path planning methods suffer from insufficient real-time performance and low collision detection accuracy when applied in underground environments. Therefore, based on the traditional RRT* algorithm, the Wd-RRT* algorithm suitable for underground mining scenarios is developed. Using the Wd-RRT* algorithm, an automatic underground parking path planning framework is constructed, which includes three processes: node generation integrated with an artificial potential field, path generation and smoothing incorporating Reeds-Shepp (RS) curves, and swept area generation along with collision detection. To better adapt to the underground mining environment, unlike traditional rectangular bounding box collision detection strategies, this paper combines the wheelbase difference with the planned path to generate the vehicle's swept area during movement, and performs collision detection based on this swept area to ensure safety and improve efficiency. The study conducted numerical simulation experiments, 1∶1 simulated tunnel field tests, and actual underground tunnel field tests, collecting a total of 180 experimental datasets. Simulation results show that compared to the Informed-RRT* algorithm, the Wd-RRT* algorithm reduces the average planning time by 28.67%, shortens the average path length by 5.76%, and decreases the average number of nodes by 3.95%, thereby better meeting the real-time requirements of underground parking path planning. Field test results indicate that the mining truck's lateral tracking error does not exceed 40 cm, the heading angle tracking error remains within 0.2 rad, and the minimum distance to obstacles is 65.32 cm. The paths planned by the Wd-RRT* algorithm exhibit smooth curvature and good trackability while satisfying the safety requirements for underground parking.