Collaborative Autonomy for Underwater Surveys Using Decentralized Hungarian Assignment

This work outlines the methodology of a decentralized autonomy framework for scalable collaboration of unmanned underwater vehicles performing line survey operations. This framework utilizes the Hungarian method in order to efficiently allocate mission survey lines among the agents, as well as create routes for the vehicles to complete their assigned survey lines. Modifications for incorporating unbalanced assignment problems, integrating heterogeneous vehicle operations, improving the computational complexity of the algorithm, and overcoming single-agent failures are also discussed. Finally, the performance of the autonomy package is assessed with respect to standard methods involving greedy assignments, as well as optimal solutions through exhaustive search.