Multiple Autonomous Underwater Vehicle Motion Planning for Passive Acoustic Monitoring

Underwater passive acoustic monitoring is gaining relevant significance thanks to the advent of marine robotics that opened up the possibility of using groups of autonomous underwater vehicles working in cooperation. Despite the tremendous potential, underwater communication is mainly acoustic, therefore intermittent, and often plagued with major difficulties. As a consequence, using distributed robotic systems underwater is still challenging, especially, concerning autonomy and cooperation between agents. In this context, we propose a distributed motion optimization strategy robust to latencies and packet losses. The objective of the optimization is to increase the performance and autonomy of the monitoring system in tracking an unknown underwater acoustic source. This work aims to show that the problem of tracking an underwater acoustic source can be cast in the form of a Partially Observable Markov Decision Problem to be solved in the context of a distributed network using the sequential multi-agent decision-making paradigm. To validate the proposed methodology, a distributed framework was developed to control a team of AUVs that play the role of acoustic sensor nodes in a multiple-vehicle network. The performance of the resulting acoustic passive monitoring system was assessed using a simulation environment that considers the major limitations associated with acoustic communications.