Data-Driven Exploration of Lentic Water Bodies with ASVs Guided by Gradient-Free Optimization/Contour Detection Algorithms

This paper presents a local-path planner for water quality monitoring involving an Autonomous Surface Vehicle (ASV). The planner determines new measuring waypoints based on the information collected so far, and on two gradient-free optimization and contour-detection algorithms. In particular, the optimization algorithm generates the locations where the variable/substance under study must be measured and use them as the waypoints of the external loop of the Guidance, Navigation and Control system of our ASV. Besides, the contour algorithm obtains useful waypoints to determine the water body locations where the variable/substance under study reaches a given value. The paper also analyzes how the approach works via progressive simulations over an ASV carefully modelled with a set of non-linear differential equations. Preliminary results suggest that the approach can be useful in real-world single-ASV water-quality monitoring missions where there is not previous knowledge of the state and location of the variable/substance under study.