Abstract
This paper describes a study on the prediction characteristics of the human operator in a manual control system. The task of the operator is to predict the future value of a random signal which is displayed as a two-dimensional waveform. The experiment was designed to answer three sets of questions:
(1) How does the human operator predict the future value of a continuous random signal?
(2) How are the prediction characteristics changed by the nature of a random signal?
(3) How precisely can the human operator predict the future value of a random signal?
Analysis reveals that the prediction characteristics of the human operator are essentially differential ones which are not pure, but with low-pass characteristics. In order to account for the characteristic behavior of the operator, a sampled-data prediction model is proposed, and the validity of the model is also checked. It is found that the human operator changes his prediction characteristics adaptively to keep the system optimal.
