An Efficient Obstacle Avoidance Approach For USV Considering The Heading Angle Constraint

The application of Unmanned Surface Vessel (USV) is gaining a lot of attraction in military, commercial, and scientific researches. However, due to various potential obstacles in the actual ocean, USV must be able to detect these obstacles in real-time, and take measures to avoid them safely. In this paper, we propose a novel method: Gradient and Node Optimal Rapidly-exploring Random Trees (GO-RRT) algorithm, which leverages the principles of APF (Artificial Potential Field) algorithm and RRT(Rapidly-exploring Random Tree) algorithm, incorporating gradient vectors within the framework of RRT algorithm while optimizing node placements. This work includes: (i) thoroughly considering the constraints in USV path planning based on the advantages of RRT; (ii) adopting an improved RRT based on APF potential function, significantly improving the efficiency of path search, and optimizing the generated path to make the actual distance of the path shorter; (iii) a dynamic obstacle detection algorithm that enables USV to prevent the initiation of obstacle avoidance algorithms in unobstructed sea conditions, thereby saving computational resources. In this work, GO-RRT can improve the quality of generated paths, and effectively reduce the consumption of computational resources. Our algorithm obtains superior performance, and achieves efficient results in simulation experiments.