Knowledge Acquisition and Modification based on Rasmussen's Intelligent Systems Model

Abstract

The purpose of this research is a knowledge acquisition and modification method for an agent based on the Rasmussen's intelligent systems model for Welfare Support Systems(WSS). The WSS are systems providing physical assistance to handicapped people. The agent is a locomotion robot. The knowledge acquisition method is ABLE(Activation Bidirectional propagation LEarning)using FAMOUS(Fuzzy Associative Memory Organizing Units Systems)to recognize gesticulated instructions, and the knowledge modification method is constructed with knowledge and sensor switches which use an evaluation function and weights. Experimental show the effectiveness of this method.