Cooperation in Surveillance Autonomous Underwater Networks: Field Experience from REPMUS22

The effective use and control of robotic networks are strongly hampered by the characteristics of the underwater environment, in particular by the limitations in the communication channel. Autonomy and cooperation capabilities are the pillars to provide networks the capability to enhance their mission and to adapt to the evolving scenario.In this paper, we report the results of the network deployed in a passive sonar surveillance experiment during the Robotics Experimentation and Prototyping MUS (REPMUS22) and Dynamic Messenger (DYMS22) trials, held in Portugal in September 2022. The network was composed of four sensing nodes, two bottom-nodes and two AUVs. We focus on how the robots exchanged bearing-only measurements (typical of passive sonars) and estimated target position in real-time. An occupancy grid mapping perception layer was developed to control the network. Specific solutions are introduced to handle multiple heterogeneous sensing platforms, and to achieve a trade-off between data fusion performance and the limitation of transmitted measurement messages (to reduce self-emitted noise).The reported results represent one of the first demonstrations, in a real-world surveillance scenario, of the effectiveness of data fusion and cooperation in an underwater multi-robot, heterogeneous, passive sonar network. They are one of the first convincing examples of how cooperation and real-time data fusion are not only effective, but also viable solutions to control a robotic network in a real-world, limited-communication, surveillance scenario.