Authors: Avary Fielding
Mentors: John Salmon
Insitution: Brigham Young University
Human-robot co-manipulation is a field of study that explores humans and robots collaboratively moving objects in various applications, including areas such as search and rescue or disaster response. The focus of this particular research is the coordination of a group of 2-3 human participants to maneuver a 55-lb table and execute specific tasks, in order to draw insights on effective strategies for humans-robots teams. Within each group, a leader was equipped with a virtual reality (VR) headset, providing them with a visual representation of the end goal. In contrast, one or multiple followers, who were unable to see the desired final position, had the responsibility of following the leader to complete the task. Data pertaining to forces, torques, and position was recorded for several iterations of 18 unique table movements to investigate group strategies and learning over time. One specific maneuver involving diagonal translation in the transverse plane (xy, with +y being to the left of the leader) became of special interest and led to the formation of two key hypotheses surrounding the evolution of human coordination strategies. First, it was hypothesized that participants would initially approach the task by translating forwards (in the x direction) and then sideways (in the y direction), before potentially optimizing their strategy in subsequent trials. Second, it was anticipated that, after successfully completing the task twice, participants would adapt to a more efficient method, involving direct diagonal translation. This research aims to answer whether the transition from sequential translations to diagonal movement is a learned behavior, developed over time as groups become more adept at the maneuver. Understanding how humans naturally tend to optimize large object manipulation can be invaluable in the realm of human-robot co-manipulation, as it provides a point of reference on which to base robot behavior, ultimately improving performance and efficiency in various co-manipulation scenarios.