Associated Hermite series based Iterative Learning Control with Experience inclusion using Local Learning Approach

In this paper, associated Hermite (AH) series based learning control (AH-ILC) scheme with experience inclusion has been proposed for the tracking control of robot manipulators including actuator dynamics. Approximation of system function has been done by orthonormal projections of Hermite polynomials. AH basis functions are orthogonal in euclidean space in both time, frequency domains, since each one is isomorphic to its Fourier transform, thus the Hermite coefficients are independent to each other. AH series is used to approximate the desired and actual trajectories of the system into finite number of Hermite coefficients. AH-ILC is designed in such a way that forces the Hermite coefficients of actual output approach to the corresponding coefficients of desired trajectory which are known constants, such that the tracking control of robot manipulator is achieved. Instead of zero initial input assumption as in most of the ILC algorithms, this paper includes the idea of using past trajectory tracking experiences on initial input selection using locally weighted learning approach for new trajectory tasks. The learning controller is based on the local input-output information. A priori structure or parameters of the system model are not required. The learning controller improves tracking performance as iteration progress and experience inclusion concept reduces initial iteration errors as well as improves the convergence of bounded error. Proposed ILC algorithm has been verified through detailed simulation studies.