Path Planning Optimization in Static Environments with High Time-Cost Traversable Regions: An Approach Based on HTCTR Virtualization and RRT*-Dijkstra Collaborative Decision-Making

Static path planning algorithms generate global paths based on completely known environmental information, making them suitable for scenarios with fixed obstacle locations and emphasizing global path optimality. However, in global path planning, besides impassable obstacles, there exists a category of traversable regions that incur a higher time cost compared to continuous roads, termed High Time-Cost Traversable Regions (HTCTRs). RRT, a sampling-based global path planning algorithm, exhibits asymptotic optimality heavily dependent on the number of sampling points. Within HTCTRs, achieving a path with a near-optimal time cost typically requires a large number of samples, significantly increasing computational complexity. To balance path quality and computational efficiency, this paper proposes a method: virtualizing HTCTRs as obstacles and incorporating Dijkstra's algorithm for local path re-planning, thereby optimizing the bypass decision for segments crossing HTCTRs based on an initial path generated by RRT. The core of this method lies in comparing the time cost of the original path segment traversing the HTCTR with that of the bypass path segment planned by Dijkstra, selecting the optimal branch with the goal of minimizing the global time cost. The simulation results demonstrate that, while ensuring path feasibility, the proposed method can achieve path quality comparable to graph-search algorithms within a computational time similar to that of RRT* with a low number of samples. Compared to the low-sampling-point RRT*, our method improves path quality by 47.72% on the 80\((\times)\)80 map and by 54.45% on the 500\((\times)\)500 map. Furthermore, it shows significant improvements in both computational efficiency and path quality when compared to high-sampling-point RRT* and graph-search methods.