Upper Bound Adaptive Learning of Neural Network for the Sliding Mode Control of Underwater Robot

In this article, the adaptive learning method of the radial basic function neural network is used on the variable structure sliding mode control of underwater robot to estimate and approach the upper bound of the uncertainty and disturbance induced by hydrodynamics. With this study method, the difficulties of establishment and resolving of precision dynamic model of underwater robot can be avoided. Based on the description and setting up of the control model, a tracking MATLAB simulation was performed and a series of tests on the yaw of underwater robot with all equipments of observation and manipulators were performed in a static water pool. The results of experiment showed that this control approach is available for the underwater robot.