Uncertainty-Driven Online Path Planning for Autonomous Bathymetry Mapping Using a Tethered System

This paper presents a novel and practical uncertainty-aware path planning approach for autonomous surface vehicles (ASVs) in bathymetric survey applications. Beginning with a coverage path planning, the proposed approach obtains an initial point cloud of the water body depth measurements. The uncertainty associated with each measurement is then quantified by propagating error ranges from the GPS position, the IMU attitude, and the sonar depth sensor caused by dynamic environmental conditions, such as waves, poor GPS zones, and general sensor noise. The uncertainty of each point is combined along with predefined uncertainties of the water edge into an uncertainty field of the whole water body using a proposed nearest neighbor method. Finally, the uncertainty field is fed into an uncertainty-aware path re-planning algorithm that favors high uncertainty areas. Compared to a simple coverage path, the presented method reduces the overall uncertainty of the bathymetry map, resulting in lower errors relative to the ground truth map.