DPF-Bi-RRT*: An Improved Path Planning Algorithm for Complex 3D Environments With Adaptive Sampling and Dual Potential Field Strategy

Abstract

This research work, we introduce a path planning algorithm called DPF-Bi-RRT* which integrates a Dual Potential Field mechanism with a Targeted Sampling Strategy to address path planning issues in complex 3D spaces. The algorithm achieves a good trade off between the global path optimization and precise local obstacle avoidance by combining dual-attraction and dual-repulsion mechanisms. The algorithm achieves effective exploration of path regions of high quality by dynamically adjusting the sampling distribution. Additionally, a biased random sampling strategy improves computational efficiency by directing sampling resources toward sections with higher promise of optimal paths, dramatically reducing computational cost. The dual potential field model lead to more flexible method in collision avoidance and can improve the precision of collision avoidance, especially in cluttered dynamical spaces. We perform comparative simulations of DPF-Bi-RRT* versus RRT*, Bi-RRT*, and APF-Bi-RRT* across three defined environments to demonstrate that DPF-Bi-RRT* results in lower average node counts, less computational time, and longer path lengths than all three. Results validate its capability of generating smooth, collision free and globally optimized paths making it especially suitable for autonomous aerial vehicle (AAV) navigation in complex 3D environments.