Vision-based neural network control for constrained robots with constraint uncertainty

Research on hybrid position and force control of robot manipulators has assumed that the structure of constraint surface is known exactly. However, in many force control applications of robot, the exact model of the constraint surface cannot be obtained. In the presence of the constraint uncertainty, it is difficult to analyse the stability of the force control system. A vision-based neural network controller is proposed for robots with uncertain kinematics, dynamics and constraint surface. The proposed adaptive neural network controller does not require the exact model and structure of the constraint surface as assumed in the literature. It is shown that stability can be achieved with the uncertainties. Simulation results are presented to illustrate the performance of the proposed controller.