Data-Driven Dynamic Modeling for a Swimming Robotic Fish

This paper proposes a data-driven dynamic modeling method for multijoint robotic fish with irregular geometric profiles and numerous heterogeneous hydrodynamic parameters. The method is composed of two main components: dynamic modeling and hydrodynamic parameter identification. In dynamic modeling, fluid forces exerted on the robotic fish are analyzed by the Morrison equation and the strip method. A dynamic model with an explicit formulation is derived, in which all terms involved in the dynamic analysis are converted to the coordinate system attached to the head. Further, the parameter identification technique is integrated into dynamic modeling, which reshapes it with data-driven feature and thereby makes it be competent to model swimming robots with complex geometric profiles and numerous heterogeneous hydrodynamic parameters. Experimental data of the swimming robotic fish are collected to identify the parameters directly. The obtained dynamic model is validated by data captured under extensive motion modes like forward swimming, varying velocity, and turning. Comparisons of simulated and experimental results demonstrate the effectiveness of the method.