An Ada-RRT-based ship collision avoidance path planning algorithm

A collision avoidance path planning model is developed based on Adaptive step Rapidly-exploring Random Tree（Ada-RRT） algorithm instead of conventional RRT. The latter may experience low sampling efficiency in complex environments. The ship domain model is introduced to guarantee safety. Virtual obstacles are generated to ensure that the planned path complies with COLREGs. The time series is introduced to establish a dynamic obstacle detection mechanism. Heuristic sampling strategy is adopted to prevent the generation of invalid nodes. The mapping function between obstacle density and RRT searching step size is established. With the linear dynamic model, the preliminary current searching step size is calculated according to the previous searching step size. The preliminary step size is optimized according to the obstacle density information. Simulation is carried out. The results show that, compared with the conventional RRT model, the ship collision avoidance path planning model using Ada-RRT algorithm performs is superior in four aspects: sampling success rate, path searching time, number of path segments and total length of path.