Connected underwater robots: Theory and practice

Reliable subsea communication is the principal technical challenge that limits the ability of underwater vehicles to collaboratively perform tasks. The subsea communication channel is extremely low bandwidth and suffers from latencies and frequent dropouts. We present fundamental advances in network theory and corresponding distributed estimation and control algorithms that enable autonomous underwater vehicles to cooperate despite infrequent communication between vehicles. We describe the practical challenges of subsea operations and how our approach distributed data fusion and decentralized control addresses these challenges. We also discuss advances in stochastic graph theory that enable us to model the sparse, time-varying network topology available underwater in a manner suitable for control design. The practical utility of our approach has been verified during field trials in which a team of autonomous underwater vehicles cooperatively localized the source of an acoustic signal.