Adaptive Sampling of Ocean Processes Using an AUV with a Gaussian Proxy Model

Abstract

This paper presents methods for building and exploiting compact spatial models on board an autonomous underwater vehicle (AUV) towards tracking suspended material plumes. The research is aiming to improve real-time monitoring of dispersal dynamics connected to marine industries such as oil and mine tailing. By exploiting in-situ information from sensors, the AUV is able to assimilate and adapt the mission capitalizing on all the information available. The spatial model is built using Gaussian process approximations and an objective function for path planning is suggested to maximize the value of the collected information.