Learning to swim in potential flow

Fish swim by undulating their bodies. These propulsive motions require coordinated shape changes of a body that interacts with its fluid environment, but the specific coordination leads to robust turning and swimming remains unclear. To address problem underwater motion planning, we propose simple model three-link fish in potential flow environment use model-free reinforcement learning for control. We arrive at optimal two tasks: desired direction towards known target. This belongs class problems geometric mechanics, as driftless dynamical systems, which allow us analyze behavior terms phases over space fish. methods are less intuitive presence drift. Here, analysis tool assessing, visualizing, interpreting control policies obtained via absence then examine robustness these drift-related perturbations. Although has no direct drift itself, it learns take advantage moderate reach