産業用ロボット制御における人間-ロボット認知結合の新手法

原文アブストラクト

This paper demonstrates the groundwork for the structure and nature of Human-Robot Cognitive Coupling.The human mind is best at associating objects, while digital devices can only compare. Successful communication between robot and human agents involves translating human thought into unique numbers that can be compared, and translating these numbers back into a form that humans can understand. A cognitive robot is one capable of interpreting its own operational data. Working from a D&D project standpoint with a telerobot, a system is designed to utilize sensor data, extract usable profiles and information from them, and then apply them as a means of self-testing for nominal/off-nominal operations. If operations are considered off-nominal, the telerobot begins a sequence of recalling and comparing previous off-nominal occurrences to determine potential corrections. In this manner, a telerobot can apply previous 'experiences' in order to learn without the need for operator input. In the cognitive telerobotics, knowledge base is an important component to store required information. In this paper, a tool-centric knowledge base with learning capability is presented. In this paper, the knowledge can be updated based on the reinforcement learning algorithm at both global and local levels and knowledge fusion by coupling with operator's cognition.