This paper describes an application of the empirical modeling of natur
al natural phenomena to the optimal self-control of an autonomous syst
em in a chaotic environment. The system consists of a network of senso
rs, a modeler a controller, a plant, and a utility estimator. The mode
ler contains a self-organizing neural network and a conditional averag
e estimator An empirical model, which incorporates the influences from
the environment, the system response and the utility, is formed in th
e modeler during training, The sensors provide signals representing th
e joint state of the environment and the system, while the utility est
imator transforms these signals into a utility signal. A vector compri
sing the joint state, the control, and the utility variable is then ut
ilized in a self-organized adaptation of prototype vectors. During ada
ptation, samples of the control variable are generated either randomly
or by a reinforcement procedure, while during application the optimal
control variable is estimated by a conditional average taken over the
prototype vectors. The control variable drives the plant, and improve
s its performance. The method is demonstrated, using as examples the o
ptimal selection of cutting depth in a chaotic manufacturing process,
the self-stabilization of a randomly influenced system, and reversing
a vehicle. (C) 1997 Elsevier Science Ltd. All rights reserved.