A Neural-Network-Based Robust Controller for Robotic Flexible Endoscope with Unknown Parameters and Uncertain Disturbance

摘要

The robotic flexible endoscope is a typical continuum robot driven by tendon-sheath, which is frequently used for identifying and treating respiratory illnesses. However, tendon-sheath actuated bending mechanism is a complex nonlinear system that suffers from unidentifiable system parameters and external disturbances, which presents significant difficulties to the precise modeling and controller design for the robotic flexible endoscope. In this article, we transformed the Eular-Lagrangian dynamics of the tendon-sheath actuated bending mechanism in the robotic flexible endoscope into a form with lumped disturbance. Then, on the basis of the RBF neural network, a robust tracking controller is developed. Finally, we demonstrated the system’s stability using Lyapunov theory and confirmed the effectiveness of the controller in the simulation environment. The simulation data showed that the proposed controller reduced the tracking error by at least 4.76% and the variance by at least 3.84% compared to the PID controller. Therefore, the designed neural-network-based control strategy can effectively mitigate the impact of unknown parameters and external disturbance while achieving high precision control performance.