A Neural Networks-Based Feedback Robust Adaptive Controller for Robots

This paper presents an app.roach to feedback adaptive motion control of robot manipulators based on neural networks. The controller includes a set of trained neural networks and an update law to adjust robot dynamics and payload uncertain parameters. The controller is robust to neural networks learning errors using a sign or saturation switching function in the control law. A global stability analysis is given, as well as simulation results to show the practical feasibility and performance for the robust adaptive controller are given.