DAPF-BI-RRT: A Bidirectional Rapidly- Exploring Random Tree Algorithm Based on Hierarchical Potential Fields and Multi- Parameter Dynamic Self-Adaptation

Path planning is a core technology that enables autonomous unmanned systems (AUS) to achieve autonomous movement in complex environments. Aiming at inherent limitations of individual path planning algorithms, this paper integrates the Artificial Potential Field (APF) and Bidirectional Rapidly-Exploring Random Tree (BI-RRT) frameworks, proposing a novel Dynamic Adaptive Hierarchical Potential Field Bidirectional Rapidly-Exploring Random Tree algorithm (DAPF-BI-RRT) based on hierarchical potential fields and multi-parameter dynamic self-adaptation. First, in the random point sampling phase, an adaptive dynamic probability bias strategy optimizes the sampling process, improving sampling efficiency and enabling effective configuration space exploration. Second, in the path expansion phase, a hierarchical potential field strategy balancing strong and weak attractive forces and an environment-adaptive step size mechanism synergistically accelerate convergence; a two-stage weighting strategy for new node generation mitigates local optima and goal unreachability. Finally, simulation experiments are conducted in a 2D environment to compare the proposed DAPF-BI-RRT algorithm with the traditional RRT, BI-RRT, APF-RRT and other benchmark algorithms in obstacle-laden environments with different complexities and narrow passage environments. The experimental results demonstrate that the DAPF-BI-RRT algorithm exhibits superior performance and robust stability, and can effectively address the challenges of efficient path planning for AUS in complex environments.