Iterative learning control of repetitive processes

This paper describes the implementation of an iterative learning control loop within a motion controller that is engaged in a repetitive process. The overall performance compared to a standard solution is improved as the tracking error is greatly reduced. However, nothing is without cost and the learning controller if left unchecked will eventually drive the system unstable. The accepted approach has been to disable the learning controller once the tracking error has been reduced to a predefined level and only re-enable it if the error exceeds acceptable limits. This paper proposes a simple method to stabilise the iterative learning controller such that it will reduce the tracking error to a constant value over a number of iterations. The proposed controller may find wide application in robotic systems.