Intelligent feature estimation, based on human's perceptual features

Abstract

The paper presents a relatively new human-computer interaction paradigm, where a human operator's perceptual actions are mimicked by the computer. In this sensor controlled system concept, we estimate a specific feature characterizing, for example, product quality, and apply intelligent analysis and an optimization to assess product quality as acquired by a human expert. The human operator contributes his/her intelligence to this man-machine interaction through learning the measurement system. An illustrative example shows how the human operator's knowledge and experience are learned by a sensor based system within a complex dough mixing optimization process in an industrial bread baking plant. The resulting sensor system acts as an intelligent feature estimator in a complex industrial process for monitoring the dynamical behavior there. The system allows easy sensor observation and makes decisions based on learning interaction with a human.