An Improved Gaussian Sampling-Based Bidirectional RRT Algorithm in 3D Path Planning for Low-Altitude Urban Environments

This paper proposes an improved Bidirectional Rapidly-Exploring Random Tree (Bid-RRT) algorithm to address path-planning challenges in three-dimensional complex environments. By introducing a map expansion optimization strategy, the traditional 0–1 obstacle map is transformed into an inflated map with enhanced safety margins, significantly improving path planning safety and search efficiency. Furthermore, the algorithm incorporates a hybrid Gaussian sampling method based on the target and start points, along with a dynamic step size adjustment strategy, ensuring intelligent distribution of the sampling space and improving the feasibility of the planned path. To enhance robustness, a time-based probability factor, and adaptive potential field guidance mechanism are designed to balance path optimization and search efficiency during exploration. Experimental results demonstrate that the proposed algorithm achieves superior path optimization, success rate, and computational efficiency in complex 3D environments. This study provides an efficient and reliable solution for 3D UAV path planning.