RULE-BASE REDUCTION FOR A FUZZY HUMAN OPERATOR PERFORMANCE MODEL

Abstract

This article presents a general procedure of reducing the number of fuzzy rules needed to perform a human-in-the-loop (HIL) design process using a virtual environment design tool. This HIL design process is created for designing an adaptive steering controller to achieve optimal vehicle maneuverability regardless of operator’s driving behaviors. In this design process, a dynamic model of an articulated off-road vehicle is implemented to determine the vehicle steering maneuverability via real-time simulation, and a virtual operator model is used to generate steer actions to guide the vehicle traveling on a predetermined path. Due to the complicated nature of steering an articulated vehicle, a high degree of granularity was required to cover all possible combinations of operating conditions. In order to meet real-time simulation requirements, a hierarchical fuzzy relations control strategy (FRCS) has been developed to reduce the size of the virtual operator’s rule-base. Using the developed hierarchy, the fuzzy steering controller could effectively incorporate the reduced size rule-base. Validation simulation showed that this hierarchical approach could reduce the size of the rule base by over 98% without affecting the performance of the virtual operator.