A Flexible-Time Iterative Learning Control Framework for Linear, Time-Based Performance Objectives

This paper describes a method for optimizing a user-defined, time-based, linear system performance index through the use of a flexible-time iterative learning control (ILC) framework. This method utilizes a two stage design wherein a point-to-point ILC procedure is conducted to improve sparse reference tracking performance, followed by a linear programming optimization that updates the system timing to minimize the time-based performance cost. A guarantee of strictly monotonic improvement of system performance cost is presented. The technique is applied to a simulated servo positioning system subject to input saturation constraints in order to minimize the time required to track a sequence of waypoints. This framework relaxes restrictions of traditional ILC techniques that require a fixed trial length to allow ILC to be applied effectively to a broader range of system objectives.