A flexible, reaction-wheel-driven fish robot: Flow sensing and flow-relative control

This paper studies flow sensing and flow-relative control for a flexible fish robot actuated by an internal reaction wheel. Two flow models are presented including a quasi-steady potential flow model and an unsteady vortex-shedding model. A recursive Bayesian filter is adopted to assimilate distributed pressure measurements and a bending-curvature measurement for flow-field estimation. The dynamic model of the reaction-wheel-driven fish robot is derived. A flow-relative control strategy for tracking swimming speed and turning rate consists of a feedforward controller designed based on the inverse steady-state turning model and a feedback controller that utilizes estimated flow information and an angular velocity measurement. Simulation results are presented to demonstrate the control design.