Intention recognition of a human commanding a mobile robot

This article discusses intention and plan recognition ideas applied to human-machine interaction. Agents capable of intention recognition of other agents can interact and cooperate effectively without much a need for communication. Dynamic Bayesian networks are used to infer for the agent intentions. A simplified example related to the motion of a mobile robot is worked out with details to clarify the proposed methodology. Further, the article discusses the difference between intention and plan recognition and shows that a reactive intention recognition approach suffices for compliant human-machine interaction. The proposed intention-recognition approach does not require the full knowledge of the environment as information gathered by typical local sensors suffices