Closing the Loop on Sensor-Based Autonomy for Swarms of Underwater Robots

Characterizing dynamic structures in the ocean mesoscale such as fronts, plumes, and currents requires multiple, simultaneous in situ measurements that are difficult and often prohibitively expensive to make. These phenomena have begun to be studied using autonomous underwater robots, which typically carry state-of-the-art inertial navigation systems to sense position and acoustic modems to transmit this position alongside the observed data. However, traditional approaches are simply not scalable in cost or in communication bandwidth to large numbers of robots working collaboratively to sample at these required spatial and temporal scales. Recent work has successfully demonstrated a viable scalable solution to navigating swarms of underwater robots using a ping from a leader to control the behavior of a group of followers (Fischell, 2019). Here we present a complementary scheme whereby these followers transmit pings that are modulated by the value of an environmental sensor such as temperature or salinity. The leader can then use an acoustic array to receive all of these pings and resolve their arrival bearings to “see” what the followers are observing without uniquely resolving their positions or motions. This decoupling of control, navigation, and sensing provides real-time information for the leader to inform adaptive sampling strategies in a way that is truly scalable to swarming behaviors in the ocean.