Multi-AUV Adaptive Path Planning and Cooperative Sampling for Ocean Scalar Field Estimation

Accurate estimation of water quality is significant for ocean resource development. The autonomous discovery and intensive sampling of high chlorophyll-a concentration area is meaningful for early warning of algal blooms. In this article, an online path planning method with heterogeneous strategies and low-communication cooperation is proposed, and it is used for multiple autonomous underwater vehicles (AUVs) adaptive sampling. The environmental scalar field is modeled as a Gaussian process (GP), which is learned from the accumulated measurements gathered by the multi-AUV system equipped with sensors. The heterogeneous strategy cooperative sampling (HSCS) method is proposed, so that every AUV can plan its own path based on the newly designed differential evolution (DE)-based path planner. Meanwhile, the sparse variation GP (SVGP) is introduced to summarize local measurements, which endows AUVs to share information effectively with weak communication environment. Simulation experiments are carried out to test the performance of the proposed algorithms, and the simulation result shows that the proposed method can reduce the estimate error by at least 15.6% with limited communication, and the coverage integrity of the interesting region is guaranteed.