Learning Motor Trajectory by Iteration for Elastic Joint Industrial Robot Manipulators

This paper proposes a process to design an iterative learning control (ILC) to be applied to flexible joint robot manipulators which repeats identical tasks under the same operation conditions. The proposed ILC updates the desired trajectory not the feedback control signals in order to avoid potentially destabilizing embedded controllers of the robot manipulators. To guarantee the convergence of tracking errors, two types of filters are introduced. One is motor encoder to get real time motor state and the other is an external sensor to collect link state for off line process. System transfer function is required during filter design process to achieve stability in the iteration domain. The estimated system transfer function is used to make sure the stability of the designed filters. Simulation is done with a 2-link flexible joint robot manipulator. The simulation shows that joint errors converge to zero as the learning process proceeds. Key–Words: joint flexibility, iterative learning control, robot manipulators, tracking