Reaction-Wheel-Based Roll Stabilization for a Robotic Fish Using Neural Network Sliding Mode Control

The intrinsically reciprocating motion in fishlike propulsion causes severe attitude instability of a robotic fish, which poses enormous challenges for environmental perception and autonomous operation. To address this issue, in this article, we propose a reaction-wheel-based control framework for guaranteeing the roll stability of the robotic fish. The mechatronic design and dynamic model of the designed robotic fish with an internal rotor are presented. By means of the simplified model and frequency domain analysis, the effect factors about roll stability are concretely analyzed. More importantly, a hybrid controller that combines a sliding mode controller with a neural network feedforward compensator is developed to reject the severe disturbance on roll angle. Then, the Lyapunov stability theory is utilized to analyze the stability and convergence property of the closed-loop system. Finally, the experimental results show that the proposed methods possess more significant performances than the passive stabilization method, which provides a valuable reference for attitude stabilization control and robust environmental perception of underwater robots.