Sunlight Algorithm: A Novel Shortest Path Planning Technique for Unmanned Surface Vehicles With Enhanced Search Efficiency

This paper investigates the shortest path planning of unmanned surface vehicles (USV) in complicated marine environments. A novel path planning method called “sunlight algorithm” is first presented by mimicking the radiation of the sun on the surface. The essence of this algorithm follows from the fact that the shortest path always occurs at the corners of obstacles. Different from the RRT* (Rapidly-exploring Random Tree Star) algorithm, which is characterized by sparse and random sampling, the proposed method only samples the tangent points of the sunlight on the obstacle edges. By iteratively treating the tangent point as a new sun, the proposed scheme can search the shortest path more efficiently. The implementation of this algorithm has four notable features: 1) a point filtering mechanism through the father-son relationship in the open set eliminates the insignificant sampling; 2) the setting of forward distances endows the planned route with the ability to fit the curve geometric shape of the obstacle; 3) the heuristic function gives the tuning between optimality and rapidity, and the probabilistic selection brings in the random exploration with robustness; 4) the bidirectional back-end processing procedure makes the further improvement on the initially planned paths. By using OpenCV simulations, it is proved that the proposed scheme outperforms the others in searching the shortest path. The open source codes are available at https://github.com/dengyingjie1993/Dengyingjie-algorithm-for-IEEE-TIS