Hexagonal Grid Graph as a Basis for Adaptive Sampling of Ocean Gradients using AUVs

In this paper we present an adaptive sampling strategy for Autonomous Underwater Vehicles (AUVs) seeking to explore and map ocean gradients using a hexagonal grid as its potential path. The laborious and deterministic method of pre-programming trajectories has gradually been replaced by intelligent and adaptive sensing strategies. These both allows more effective use of sampling resources, as well as improved spatiotemporal resolution. We present a method for adaptive and data-driven sampling in the water-column with AUVs that follows the sense-plan-act control paradigm. The method uses a hexagonal grid graph to discretize the survey area into a honeycomb-pattern, which gives both equilateral survey paths and a small branching factor. Using this graph, we present results from a typical environmental sampling application where the goal is to search for and explore temperature gradients. The main use case for this algorithm is to prioritize the strongest gradients within the AUV spatiotemporal envelope. We present results from field trials in Trondheimsfjorden, Norway, where the AUV successfully explored a river front. We find that the algorithm performs as expected, exploring the area and revisiting sections containing gradients.