Semiotic Approach to Perceptual State Construction for Behavior-Based Robot through Recursive and Progressive Deepening Utilization of Memory
Tetsuo Sawaragi* and Satoshi Iwatsu**
*Department of Precision Engineering, Graduate School of Engineering, Kyoto University, Yoshida Honmachi, Sakyo, Kyoto 606-8501, Japan
**Industrial Electronics and Systems Lab., Mitsubishi Electric Corp., Tsukaguchi-Honmachi, Amagasaki, Hyogo 661-8661, Japan
The sources of intelligence for primitively facilitated robotic agents exist in their reconceptualizing capability of what they physically sense from the environment. To implement this capability, we adopt an inductive learning method of conceptual formation for adaptively organizing a state space and develop a new algorithm for a robot to construct its task-relevant state space efficiently through matching encountering states with the similar situations in the past and through generalizing them. We propose a methodology to dynamically increase the resolution of state spaces both adaptively and selectively by applying a concept formation technique from machine learning recursively to a record of a sensorimotor history of a learning agent. By connecting this with conventional reinforcement learning, we showed our algorithm can perform tasks without suffering from hidden state problems in an artificial maze environment, and also present its robustness even for a robot whose perceptual resources are quite restricted and/or bounded.
This article is published under a Creative Commons Attribution-NoDerivatives 4.0 International License.
Copyright© 2000 by Fuji Technology Press Ltd. and Japan Society of Mechanical Engineers. All right reserved.