In order to be effective teammates, robots need to be able to understandhigh-level human behavior to recognize, anticipate, and adapt to human motion.We have designed a new approach to enable robots to perceive human group motionin real-time, anticipate future actions, and synthesize their own motionaccordingly. We explore this within the context of joint action, where humansand robots move together synchronously. In this paper, we present ananticipation method which takes high-level group behavior into account. Wevalidate the method within a human-robot interaction scenario, where anautonomous mobile robot observes a team of human dancers, and then successfullyand contingently coordinates its movements to "join the dance". We compared theresults of our anticipation method to move the robot with another method whichdid not rely on high-level group behavior, and found our method performedbetter both in terms of more closely synchronizing the robot's motion to theteam, and also exhibiting more contingent and fluent motion. These findingssuggest that the robot performs better when it has an understanding ofhigh-level group behavior than when it does not. This work will help enableothers in the robotics community to build more fluent and adaptable robots inthe future.
展开▼