Inspection Path Planning for UAV in Substation Areas Based on The Improved RRT Algorithm

In this paper, an inspection path planning for unmanned aerial vehicle is developed based on the Rapidly-exploring Random Tree (RRT) algorithm. Firstly, based on the spatial structural characteristics of the substation, the substation is divided into regions according to reasonable area division principles, and an improved RRT algorithm is introduced for planning inspection paths within these regions. The improved RRT algorithm reduces the generation of redundant nodes and enhances the global optimality of the path by incorporating Random Geometric Graphs (RGG), pruning optimization strategies, and gradient descent-based convex optimization methods, effectively lowering computational costs. Finally, B-spline curves are used to process the smoothness of the path, making the generated flight routes more suitable for actual flight conditions. Simulation results indicate that the proposed algorithm can quickly generate safe, feasible, and optimal inspection paths on a three-dimensional occupancy grid compared to the original RRT algorithm, with a 6% increase in path distance, a reduction of 56 in the number of path grids, and a decrease of 844 in the total number of searched grids, significantly improving the efficiency and adaptability of drone inspections.