Path Planning Method Based on Artificial Potential Field and Reinforcement Learning for Intervention AUVs

In order to realize an intervention autonomous underwater vehicle (I-AUV) to remove sea urchins at affordable cost, a path planning method is proposed. The method consists of two parts. First one is seafloor tracking to observe seafloor. The second one is catching sea urchins. In the first part, the path is produced based on limited environmental information available by low-cost sensors in unknown environment. An artificial potential field based on binary Bayes filter using measurements of a mechanical scanning imaging sonar is used. The method has high real-time performance. The method was verified in the experiment, in which an AUV succeeded in tracking vertical walls keeping the reference distance of 2 m. In the second part, the path is produced based on reinforcement learning in a simulated environment. In the simulation, the agent succeeded in finding a safe path to catch sea urchins in a complex situation.