Quad Rapidly-Exploring Random Tree Star Algorithm With Improved Potential Force for Unmanned Aerial Vehicle Path Planning

Rapidly-exploring random tree star (RRT*) has attracted intensive attention in track planning due to its asymptotic optimal properties. However, the RRT* algorithm plans costly trajectory paths. Quad-RRT* algorithm with improved potential force for unmanned aerial vehicle (UAV) path planning is presented for reducing the track length of RRT* in this thesis. Firstly, the algorithm generates four random trees simultaneously in map, and it generates two random trees respectively at the starting point and the target point to improve the search efficiency. In the sampling stage of the algorithm, the random points are generated in a fixed region restricted by the sampling equation constraint method, which improves the node quality and decreases the track cost. Secondly, the potential attraction to the new node is added into the algorithm to improve new nodes quality and shorten the convergence time. Finally, the novel method is compared with other algorithms in complex environments, which proves the novel algorithm has good performance in the number of nodes, convergence time and track length.