Multi-agent system for a human-friendly robot

Abstract

At the Center for Intelligent Systems, we are working on human robot interaction (HRI). We present our approach for HRI. Our goals for HRI are to allow the human to influence the robot's actions in some way and acquire feedback about the robot's internal state. Additionally, we would like the robot to confirm the human's input and request the human's assistance if necessary. To achieve our goals, we are developing two intelligent agents: the human agent and the self agent. The human agent encapsulates what the robot knows about the human. The self agent maintains knowledge about the internal state of the robot and communicates with the human. We performed a preliminary demonstration of the system, in which the human could have limited conversation with the robot, a dual-armed humanoid. This limited conversation included commands to the robot and queries about the robot's abilities and internal state.