Development and Verification of a Gesture-generating Architecture for Conversational Humanoid Robots

摘要

With the increasing need for care and daily life support, particularly for elderly people, the demand for communication robots is steadily increasing. A robot should possess excellent conversational capabilities to coexist with humans. Gestures accompanying spoken sentences are also an important factor to consider. Humanoid robots should essentially possess such capabilities because their merits reside in their humanlike appearance that enables the use of gestures to enrich interactions. However, present systems for gesture generation lack some degree of physical realism. A method for humanoid robots to generate gestures along with spoken sentences is proposed herein. The results of a user study show that our method, which considers speech semantics, can enable robots to interact more naturally with humans. In addition, we emphasize that our gesture-generating architecture can be applied to any type of humanoid robot because our gesture presentation uses Labanotation, which is a system for describing human gestures that are easily compatible with other robots. This study will possibly contribute to the development of an integrated system for conversational robots whose speech and gestures can supplement each other in human-robot interactions.