Abstract
It seems that the emergence of intelligence in an autonomous robot exists in the dexterity of human or creatures as complex systems and the research style and the development procedure along this approach should be necessary for realization of a real intelligent robot. On the other hand, since the severe judgment of situation is required during stabilizing control of an unstable system like an inverted pendulum on a cart by human operators, it can be expected that human operators exhibit a complex behavior occasionally. This paper tries to acquire the skill of human operator and investigates the possibility of the formation of complex system in the learning process of human operators withdifficult control objects. It also considers the mechanism of robustness of human operator against the disturbance. The identified neural network as a controller from time series data of each trial of each operator show well the human-generated decision-making characteristics with the chaos and the large amount of disorder, further showing that the estimated degree of freedom of motion increases and the estimated amount of disorder decreases with an increase of proficiency.
