Spatially perceptual path planning for AUVs in uncharted marine environments using rolling window strategy
Wenlong Meng, Yanbo Pu, Yujing Li, Jinglin Wang, Gong Ya · Ocean Engineering · 2024
Autonomous underwater vehicles (AUVs) are increasingly integral to the in-depth exploration of oceanic environments. Efficiently identifying optimal routes, particularly within uncharted marine domains, is paramount for AUVs when addressing path-planning challenges. Existing path planning algorithms relying on underwater sensing equipment frequently encounter challenges in achieving efficient exploration while generating high-quality trajectories. In this paper, we introduce an innovative method aimed at efficiently generating optimal paths for AUV navigation within the constraints of limited information. In detail, given the uncertain marine environment, we transform the global one-time path planning problem into multiple iterations of local planning through a rolling window strategy. Within each successive rolling window, utilizing sensor-detected environmental data, we present a novel variant of the RRT* algorithm denoted as circle-RRT*. This algorithm is designed to determine the high-precision sub-paths efficiently. In the circle-RRT* procedure, the integration of adaptive circle sampling and the rewiring process from traditional RRT* leads to a notable reduction in redundant sampling points and modifies the connectivity relationships of the RRT tree. Consequently, these improvements enhance the efficiency of local path planning, concurrently leading to a substantial decrease in path length. We rigorously assessed the performance of our algorithm in uncharted environments via extensive simulations and conducted comparative analyses against established state-of-the-art methods. • Rolling planning boosts accuracy by dividing trajectories into smaller parts. • Sub-target selection minimizes potential energy in the force field. potential energy in the force field. • Circle-RRT* combines circle-RRT efficiency with RRT* optimality.