A human interface for an intelligent mobile robot

Abstract

We propose that using the robot control based on situation assessment may lead to the realization of a truly user-friendly human-robot interface. Two aspects of user friendliness are addressed: ease of introducing domain-specific knowledge into the robot's memory in preparation of its actual operation, and ease of giving orders to the robot. Two abilities of the robot that result from the situation-oriented approach are key to this user friendliness: its ability to acquire knowledge regarding the environment largely by learning, instead of depending on explicit inputs by an operator; and its ability to communicate with the operator on a common basis of the situation being perceived by both. These abilities, together with a specific structure of the robot's internal knowledge representation, allow the operator to communicate with the robot in a similar way as he/she would with a human whom he/she requests, for instance, to do an errand.