AUR serves as a research platform investigating notions of fluency, embodiment, and nonverbal behavior. While two humans performing an activity together quickly arrive at a high level of coordination and adaptation, in particular when they are well-accustomed to the task and to each other. This research tries to achieve a similar quality of collaboration between a human and a robot.

We support a complete-system, embodied approach to robot cognition. Following a host of neuro-cognitive research supporting a perceptual-symbol approach, we have developed an perceptual-symbol-based cognitive architecture for fluent human-robot interaction, using perceptual simulation and anticipatory emulation.

Through it, the robot learns from repetitive practice, and increasingly coordinates its actions with those of a human partner through the alteration of perceptual bias during motor action.

In studies, we found significant differences in the efficiency and fluency of the teamwork, when comparing our architecture to a purely reactive robot with similar capabilities. We also found significant differences the attitude of the human subjects towards the robot, rating it as more intelligent, contributing more to the team, and understanding the human's goals better. There were interesting differences in the language used to describe the robot. For example, only when the robot used our fluency mechanisms, did people attribute gender to the robot, they rated the robot more positively, but also tended more towards self-deprecation.