A dynamically hybrid path planning for unmanned surface vehicles based on non-uniform Theta* and improved dynamic windows approach

This paper presents a dynamically hybrid path planning scheme incorporating the global guidance, realtime motion planning and collision-free maneuvering for Unmanned Surface Vehicles (USVs). To ensure safe navigation in narrow waters, the non-uniform costmaps with risk function is designed for global planning. Non-uniform Theta* (NT*) performs a reverse search from the goal node to establish the mapping relation containing the parent node and minimum path-cost of each cell. To prevent USV trapping into local minimum after avoiding the dynamic obstacles, the parent node is dynamically selected as the local target of improved dynamic window approach (IDWA), rather than tracing the sub-target fixed in global path as DWA does. In order to intensify navigation efficiency of the USV, the minimum path-cost from current cell to the goal is also implemented to the objective function of IDWA for evaluating the predicted trajectories. The proposed algorithms are validated by comparative numerical analysis and proven to work effectively.