Position-based Dubins-RRT* path planning algorithm for autonomous surface vehicles

Path planning is a critical technology for developing autonomous surface vehicles (ASVs). The goal is to plan a collision-free, feasible, and optimal path from a defined start state to a target state in an environment with obstacles. The existing Dubins-RRT* algorithm can satisfy this requirement, but it is still inadequate in terms of path quality and computational efficiency. This paper proposes an efficient position-based Dubins-RRT* (P-Dubins-RRT*) algorithm. Specifically, the local Dubins problem is reduced to a relaxed Dubins problem, as only the position coordinates are sampled during the planning process. The fundamental expansion curves are shortened Dubins paths, while complete Dubins paths are employed for goal-state connection and rewire. The efficiency of path planning is enhanced by accelerating tree expansion, which is achieved by eliminating the constraint of heading angle. Simulation results indicate that the proposed P-Dubins-RRT* efficiently balances planning efficiency and path feasibility. On the test maps, P-Dubins-RRT* reduces the initial path generation time by an average of 70.3% compared to Dubins-RRT* and demonstrates superiority in path quality. The proposed method is applicable to issues of local path planning for vehicles with turning constraints in two-dimensional space.