Harnessing the nonrepetitiveness in iterative learning control