Generalization of feedback error learning (FEL) to strictly proper MIMO systems

This paper studies MIMO (multi input multi output) learning control systems. It generalizes the well-known feedback error learning (FEL) scheme of Kawato et al. to MIMO systems which are not necessarily biproper, and hence not invertible. Assuming that a "diagonal interactor" is known, ideal and adaptive feedforward controllers are derived to achieve FEL. It is also proven that the stability of the generalized scheme is guaranteed under a positive realness condition. Simulation results illustrate the effectiveness of the proposed method