An Adaptive Bounded Hybrid A* Planner for Autonomous Mining Trucks in Confined Unstructured Mining Environments

Automated valet-parking trajectory planning has been widely studied in structured environments, whereas its counterpart in unstructured environments remains insufficiently explored. The working face of an open-pit mine is a typical unstructured production environment, where autonomous mining trucks (AMTs) must perform collision-free and kinematically feasible maneuvers under irregular boundaries, locally confined passages, large vehicle footprints, and strict terminal pose requirements. To address this problem, we introduce an adaptive bounded Hybrid A* planner for AMT maneuver planning at mining working faces. The proposed planner integrates local-geometry-aware top-K motion primitive scheduling, adaptive rollout-length assignment, progress-reward-based dual-cost correction, and bounded continuation after the first feasible Reeds–Shepp closure. By ranking and filtering motion primitives according to obstacle clearance, maneuver margin, goal distance, and heading alignment, the planner reduces ineffective successor generation during search. By adapting rollout lengths to local spatial constraints and primitive curvature, it improves the consistency between expansion scale and environmental structure. By retaining the first feasible Reeds–Shepp connection as an initial upper bound and continuing bounded refinement, it improves the final returned path beyond the first feasible solution. Comparative studies demonstrate that, relative to the conventional Hybrid A* baseline, the proposed planner reduces the average final planning time, generated nodes, and final path cost by 54.73%, 74.10%, and 12.76%, respectively, in the abstract mine-like scenarios. In the real-map-derived mining working-face scenarios, the corresponding reductions are 25.12%, 37.98%, and 7.15%, respectively. These results indicate that the proposed method reduces search redundancy while improving the efficiency–quality trade-off in confined and geometrically irregular working-face environments.