Learning and adaptation in control systems with interactive computer graphics

A neural-based computer-aided control system design tool is described. For control purposes, a learning and adaptive neural controller, called the inverse-dynamics neural controller, is proposed for unknown linear dynamical plants. The mathematical relationship describing the dynamics of the plant under control does not need to be known a priori, and no teaching signal is required to guide the learning and control actions. An interactive computer graphic tool developed to study and monitor the performance of the proposed control scheme is described. This package is intended for research and education purposes