An algorithm for iterative learning control is developed on the basis
of an optimization principle which has been used previously to derive
gradient-type algorithms. The new algorithm has numerous benefits whic
h include realization in terms of Riccati feedback and feedforward com
ponents. This realization also has the advantage of implicitly ensurin
g automatic step size selection and hence guaranteeing convergence wit
hout the need for empirical choice of parameters. The algorithm is exp
ressed as a very general norm optimization problem in a Hilbert space
setting and hence, in principle, can be used for both continuous and d
iscrete time systems. A basic relationship with almost singular optima
l control is outlined. The theoretical results are illustrated by simu
lation studies which highlight the dependence of the speed of converge
nce on parameters chosen to represent the norm of the signals appearin
g in the optimization problem.