Grad Seminar: Adaptive Task Selection and Planning in Human-Robot Collaboration: To Be a Leader or a Follower?

Abstract

The emergence of collaborative robots has enabled robots to work closely with humans and share their workspace with them, rather than being confined in cages. Although these robots are slower and less powerful, human-robot complementary skills lay the groundwork for productive, flexible, and profitable human-robot collaboration. To exploit these capabilities and establish a seamless collaboration, human-aware programming of the robot is essential. Through this, the robot can learn, adapt, and work robustly. Most studies on this topic are primarily centered on enhancing human satisfaction and perception of the robot, as a high level of human perception of the robot aids in effective long-term collaboration. However, a question remains: "Does only the human teammate's satisfaction and perception matter?" Cobots are less expensive than traditional industrial robots, but they still need to demonstrate strong productivity to attract business investment. This underscores the importance of high performance in human-robot collaboration to ensure a positive return on investment. Consequently, effective cobot programming should address both team performance and human perception of the robot and collaboration. The robot's high computational and planning abilities allow it to fit into the role of leader in this collaboration and planning for the team to ensure high team performance. However, the human preference and the robot's (optimal) plan do not always align. This problem necessitates cobot programming to enable them to adjust to the preferences of their human coworkers, thereby upholding a high level of satisfaction and perception while maintaining acceptable team performance. This seminar focuses on the proposed framework and algorithms designed to estimate human preferences and performance and incorporate them into the robot's task selection and planning. Furthermore, the effectiveness of the proposed framework will be examined through the results obtained from simulation studies and its implementation on an actual robot.

Presenter

Ali Asl, PhD candidate in Systems Design Engineering 

Attending this seminar will count towards the graduate student seminar attendance milestone!

Join in person E5-6127 or on Teams