Qian,Rui (Brigham Young University)
Faculty Advisor: Salmon, John (BYU - Ira A. Fulton College of Engineering, Mechanical Engineering); Killpack, Marc (BYU - Ira A. Fulton College of Engineering, Mechanical Engineering)
The cooperation between humans and robots may become more intuitive as technology develops. It is foreseeable that soon physical human-robot collaboration may be applied in the area of co-manipulation of objects, especially in search and rescue. It comes naturally for a human dyad to adapt and respond to changes with each other while moving objects. However, it still can be difficult for a robot to determine the motion it should take to best collaborate with a human. In order to optimize robots imitation of humans and improve their efficiency to assist humans, the research aims to design algorithms for robots to move objects in more human-like ways by first analyzing behavioral characteristics of human-to-human collaborations.
During our experiments, we will designate one person per group as a leader and one as a follower to carry a stretcher-like table as a simulated object with force-torque sensors through different obstacles. As the follower will not be explicitly told the intention of the leader, the forces and torques that the follower feels through the object become important for understanding the leader's intent. With standardized specific goals and qualifiers, data will be gathered on the force and torque people exert on the object and motion of table; we will then analyze the correlation and characteristics between the data and people's actual intentions. The data will later be implemented as an algorithm on the robot to help it identify human's intentions and to complete the cooperative task efficiently and smoothly.