Persistent Autonomy for Aquatic Robotics: the Role of Control and Learning in Single and Multi-Robot Systems

There has been a steady increase in the deployment of autonomous aquatic vehicles for applications such as hazardous waste mitigation, inspection and recovery of underwater structures, environmental monitoring, and tracking of ocean processes. These emerging applications require solving unique challenges that arise in the underwater and aquatic environments, as well as realizing a persistent presence in the environment. The lack of reliable wireless communications underwater makes remote control difficult; aquatic vehicle dynamics are tightly coupled with environmental dynamics making controls hard; and well-understood perception technologies do not always apply to the aquatic environment. These challenges, in addition to our limited understanding of the complexities of the fluidic environment, make closed-loop control, online learning, and adaptive decision making challenging at best. Such challenges are further complicated when considering long-term autonomy, a multi-robot system, or both. Here, we unite experts in the interdisciplinary field of autonomous aquatic robotics to bridge the gap between (1) modeling and prediction for persistent, closed-loop control, (2) online learning in highly dynamic and uncertain environments, and (3) coordination of heterogeneous multi-robot teams. Specifically we highlight new work that lies at the intersection of robotics, control theory, artificial intelligence, machine learning, ocean science, long-term autonomy and transport theory.