“…The critical property of such models is the fact that, although trained on sample points from a single trajectory, the network is capable of accurate generalization on other, hitherto unseen, trajectories. Building on the effectiveness of short range prediction using neuro-models, we recently proposed an adaptive neuro-genetic control architecture, a hybrid method of adaptive control using neural networks for modelling and genetic algorithms for control [2]. The neuro-genetic control system was demonstrated with reference to the attitude control problem for a rigid body (satellite) equipped with thrusters about each principal axis in the case of large slew angles, unknown dynamic characteristics, and stable and unstable target states.…”