Tracking control and neural network disturbance observer for a 6 DoF underwater welding robot

Abstract

This article addresses the control of an underwater welding robot in 3D space. A control approach combining a finite time sliding mode controller for position and orientation control and feedback linearization for controlling one degree of freedom of the welding arm is adopted. The utilization of underwater robots in marine environments is often hindered by uncertainties and disturbances induced by ocean waves and currents, resulting in decreased accuracy and operational disruptions. To overcome these challenges, a novel observer based on a radial basis function neural network is developed to enhance the performance of the underwater welding robot. The neural network’s weights are optimized using the Lyapunov method within the control law framework. Through simulations, the article evaluates the observer’s efficacy in accurately tracking reference trajectories in the presence of uncertainties. The results underscore the significant contribution of this estimator in mitigating uncertainties and disturbances, thereby substantially improving the overall performance and operational reliability of underwater welding robots. The control strategies and observer design presented in this study pave the way for enhanced accuracy, stability, and efficiency in complex underwater welding operations.