Study on turning mechanism of robotic fish with given radius

Abstract

• Analyze the characteristics of fish body turning motion, increase the bias of flexible fish body, and provide a kinematic model of fish body flexible turning motion, laying a theoretical foundation for the realization of turning motion in robotic fish. • Based on the established kinematic model, select the offset of the flexible fish body, the time to bend to the maximum moment, and the oscillation frequency of the robot’s tail fin as inputs. Establish an RBF neural network by using the average turning speed, turning time, and angular velocity as outputs. • Inverse solve the neural network to achieve the rotational motion of the robotic fish within a specified time. It provides theoretical support for precise control of the turning motion of robotic fish. In this study, we aim to develop a kinematic model with fixed radius turning for robotic fish in complex environments. To this end, we adjusted the fish body wave equation and proposed the offset of a flexible fish body, the time of bending to the maximum moment, and the frequency of swing. Additionally, we employed computational fluid dynamics (CFD) to analyze the relationship between the offset of the flexible fish body, time of bending to the maximum moment and frequency of swing, and the average turning speed, turning time, and angular speed, which was used as a dataset. The dataset served as the inputs and outputs for the radial basis function(RBF) neural network to develop a dynamic model of the robotic fish’s turning motion. Subsequently, we calculated the turning radius of the robotic fish based on the turning kinematics. Based on the turning time and radius of the robotic fish, the inverse solution of the dynamic model can be used to achieve the rotational motion of the robotic fish within a specified time. Additionally, we conducted physical experiments to verify the established dynamic model using the existing robotic fish in the laboratory. The results indicate that the pectoral and caudal fins periodically generated jets and vortices during the turning process, jointly driving the turning and swimming of the robotic fish. When the pectoral and caudal fins swing, a significant pressure difference occurs on both sides, and the high-pressure vortex sheds backward to provide thrust in the forward direction. The results of the corresponding experiment demonstrated that the average tracking error was 6.4 %. Therefore, the proposed method ensures that the robotic fish can swim based on the given radius, which is essential for its precise motion control.