Surfacing strategies for multi-robot adaptive informative sampling with a surface-based data hub

Autonomous underwater vehicles (AUVs) are increasingly being used for lake and ocean sampling. Adaptive informative sampling is one method to gather data that is most informative for a phenomena being modeled. An open question is how multiple AUVs can best collaborate in collecting this data. In this work, we propose two new approaches for deciding when to surface for data sharing with a surface-based data hub: altruistic surfacing, based on the amount of information a vehicle has to share, and gain-based surfacing, based on an estimate of how much information is available at the surface. We compare these approaches to three prior approaches: timed surfacing - with and without data hub - and synchronous surfacing with Voronoi-based partitioning for coordination. Our simulation results show that all approaches perform approximately equal in terms of the error between models created by the vehicles and the ground truth, evaluated over time. While the Voronoi-based approach performs has wider variance in its initial performance, it can perform better at later sampling times for a subset of the tested scenarios. The altruistic surfacing approach can greatly reduce the number of surfacing events required, and shows promise for sampling at greater depths.