Oral Candidacy - Tariq Iqbal
|Start:||4/6/2016 at 2:30PM|
|End:||4/6/2016 at 5:00PM|
|Location:||117I & J Cushing Hall|
Faculty and students are welcome to attend the presentation portion of the defense.
April 6, 2016
117I & J Cushing Hall
Adviser: Dr. Laurel Riek
Committee Members: Dr. Patrick Flynn, Dr. Michael Richardson, Dr. Walter Scheirer, Dr. Julie A. Shah
Coordination Dynamics in Human Robot Teams
As technology advances, the presence of autonomous robots is becoming prominent in our daily life, and they are expected to interact with humans in teams. In order to be effective teammates, robots need to be able to understand human team dynamics. This understanding will help the robots to recognize, anticipate, and adapt to human motion. Understanding human team dynamics is not trivial for robots. One of the crucial reasons for that is the limited number of available approaches for robots to detect human team dynamics accurately.
In many group interaction scenarios, humans coordinate their activities with the rest of the group, and the group eventually reaches a synchronous state. How synchronously the group members perform their actions is an important behavioral indicator of group-level cohesiveness, and also important for accurately understanding the affective behavior of a group. Thus, if a robot has an understanding of these underlying group dynamics, then it can anticipate future actions in a team, and adapt to those actions in order to be effective teammates.
To enable a robot to understand team dynamics, we have developed a method to automatically detect the degree of synchronization of a group. This method takes multiple types of discrete, task-level events into consideration while measuring the group synchrony. In addition, this method is capable of modeling synchronous and asynchronous behavior of the group. We have explored this method within the context of joint action, and have validated the method by applying it to a human-human, and a human-robot team. Our results suggest that our method is more accurate in estimating synchronization of a group than other methods from the literature that depend on a single event type.
Based on this work, we have designed a new approach to enable robots to perceive human group behavior in real-time, anticipate future actions, and synthesize their own motion accordingly. We validated this method within a human-robot interaction scenario, where an autonomous mobile robot observes a team of human dancers, and then successfully and contingently coordinates its movements to ``join the dance''. We compared the results of our anticipation method to move the robot with another method which did not rely on team dynamics, and found our method performed better both in terms of more closely synchronizing the robot's motion to the team, and also in exhibiting more contingent and fluent motion. These findings suggest that the robot performs better when it has an understanding of team dynamics than when it does not.
Furthermore, we investigated the effects on the group synchronization by adding one, or more robots to a multi-human, multi-robot team. Our results suggest that heterogeneous behavior of robots in a multi-human group can play a major role in how group coordination dynamics stabilize (or fail).
Beyond the rhythmic synchronous group interactions scenarios, there exist many group interaction scenarios where the activities performed by the group members are not only synchronous, but also change their tempo over time. In my proposed work, I plan to push the current research further by exploring how robots can take advantage of the understanding of temporal adaptation and anticipation to coordinate with other external rhythms. I plan to develop models for robots to better understand the tempo changing behavior in teams, so that the robots can leverage this knowledge to coordinate with group members. I also plan to perform experimental validations to evaluate how these models affect human-robot team coordination.
My current and proposed work will enable robots to recognize, anticipate, and adapt to human groups. This work will help enable others in the robotics community to build more fluent and adaptable robots in the future, and provide a necessary understanding for how we design future human-robot teams.