Model identification for the yaw motion of a tail-actuated robotic fish

Abstract

System model is fundamental for accurate motion control of the robot. However, due to the complexity of aquatic environments and the special fin-driven structure, it is difficult to dynamically model the system of robotic fish. This paper proposes an experimental modeling method for yaw motion of a freely swimming robotic fish through system identification. Tail deflection and the corresponding angular velocity of the robot are obtained by a series of open-loop experiments. Both notch filter and low-pass filter are adopted to get rid of the oscillation disturbance in the robot angular velocity. By simplifying the robot model as a linear system, transfer functions of robotic fish are derived by data-driven model identification method. Further, yawing PID controllers are designed according to the obtained models. Finally, simulations and experimental results for the yawing control of the robot are compared, demonstrating the effectiveness of the proposed method.