Sensor based knowledge acquisition and modification
Tomomi Hashimoto, Toru Yamaguchi · 2002
This paper proposes a knowledge acquisition and modification method based on the Rasmussen model for an agent. The agent is a locomotion robot. The knowledge acquisition method uses metacognition as an estimation function, and the knowledge modification method is constructed with knowledge and sensor switches that use an evaluation function and weights. These knowledge processings operate in parallel. The hierarchical knowledge construction is an engineering model of a brain. Experimental results show the effectiveness of this method.