Maximum Information Coverage and Monitoring Path Planning with Unmanned Surface Vehicles Using Deep Reinforcement Learning

Manual monitoring large water reservoirs is a complex and high-cost task that requires many human resources. By using Autonomous Surface Vehicles, informative missions for modeling and supervising can be performed efficiently. Given a model of the uncertainty of the measurements, the minimization of entropy is proven to be a suitable criterion to find a surrogate model of the contamination map, also with complete coverage pathplanning. This work uses Proximal Policy Optimization, a Deep Reinforcement Learning algorithm, to find a suitable policy that solves this maximum information coverage path planning, whereas the obstacles are avoided. The results show that the proposed framework outperforms other methods in the literature by 32% in entropy minimization and by 63% in model accuracy.