Path planning and optimization for AUV with dual-mode sampling and dynamic step-size in coupled current-terrain environments

Objectives This paper proposes an Enhanced Bidirectional Rapidly-exploring Random Tree Star (EB - RRT*) adaptive motion planning method to address the path planning problem for Autonomous Underwater Vehicles (AUVs) operating in environments characterized by strong dynamic ocean currents and complex seabed topography. The objective is to improve the efficiency of path searching and the quality of generated paths in such challenging underwater settings. Methods The proposed method employs a bidirectional search strategy, constructing two trees simultaneously from the start point and the target point. A composite heuristic sampling mechanism is designed to integrate ocean current dynamics with seabed topographic information, enhancing the search guidance within the three-dimensional space. Furthermore, a dynamic step-size adjustment strategy coupled with the AUV's motion state is introduced, enabling the tree's expansion to adaptively optimize based on environmental constraints and the vehicle's dynamic characteristics. After generating an initial path, Non-Uniform Rational B-Splines (NURBS) are applied for smoothing. This is followed by a trajectory optimization method based on linear programming to generate a time-energy integrated optimal trajectory that satisfies kinematic constraints.Results Simulation results demonstrate that the proposed method can reliably generate safe and smooth feasible paths under the combined interference of dynamic currents and complex terrain. Compared to the A algorithm, the planned path length is reduced by 6.4%, energy consumption is decreased by 12.8%, and the average path curvature is lowered by 62.5%. When compared with RRT series algorithms, potential field-guided RRT, and other flow field-aware algorithms, the proposed method exhibits significant advantages in key performance indicators such as convergence speed, path quality, and energy efficiency.Conclusions The EB - RRT* algorithm, by effectively integrating prior environmental information with vehicle state, enables efficient and high-quality path planning in complex underwater environments, demonstrating strong potential for engineering applications.