Path Planning for Bathymetry-aided Underwater Navigation
Gao Rui, Mandar Anil Chitre · 2018 IEEE/OES Autonomous Underwater Vehicle Workshop (AUV) · 2018
It has been shown that significant variation in the bottom topography of underwater terrain could be used to dramatically improve underwater localization accuracy. However, path planning with bathymetric aids has not been extensively explored in literature. Localization accuracy strongly depends on the path taken by an underwater vehicle. Given a starting point and a destination, we develop an algorithm to plan a path that will ensure good localization. We adopt an information entropy measure to assess the localization uncertainty of a particle filter, such that a generic posterior can be described. We use this to drive a path planning algorithm that minimizes uncertainty using reinforcement learning and Gaussian process regression. We test our algorithm using bathymetric data and show that it generates near-optimal paths with good localization accuracy at the destination.