Adaptive cleaning of oil spills by autonomous vehicles under partial information

Oil spill pollution is a critical issue because of its severe influence on our economy and ecological environment. A large amount of efforts have been made for detection, monitoring and localization of oil spills; however, the cleaning of oil spills remains a challenging problem. Due to dynamic currents on the ocean surface and the limitations of the localization systems, it is not feasible to obtain the exact information of the oil spill location. Therefore, the oil spill region is approximated by an area which may not match the shape of the oil spill. Thus, an effective complete oil spill cleaning algorithm is needed that allows the autonomous vehicle to adapt its trajectories to the oil spill shape. This paper presents a concept of multi-resolution navigation for oil spill clean-up using an autonomous vehicle. The exact a prior knowledge of the oil region is assumed to be unknown or only partially known. This algorithm can dynamically discover and adaptively clean the oil spill in the workspace. A concept of multi-resolution navigation is developed which integrates local navigation and global navigation to avoid being stuck into a local minima that is generally encountered by potential field based methods. The algorithm provides complete coverage and cleaning of the oil spill via minimum overlapping trajectories, and its efficiency is validated on the high-fidelity Player/Stage simulator of mobile robots.