This work focuses on predicting the motion of a human operator by using different sensing modalities. For this project, the lab has been collaborating with Honda R&D, Japan. The work focuses on using bioelectrical signals like EMG and other sensing modalities to continuously predict human joint mechanics. This information can then be relayed as a control signal for robot teleoperation.

This work is inspired by the idea of shared control where the human operator controls a robot remotely through varying levels of autonomy. For instance, a robotic arm can track the trajectory of the human limbs and detect the intent of an operator to perform an action. However low-level tasks like grasping can be achieved autonomously after intent detection by relying on the sensory feedback information received by the robot from the environment.

The high-level goal of the project is to reduce the latency in robot teleoperation to facilitate a better human-robot interaction experience.

This project is currently under non-disclosure agreement between UMass and Honda. Please check in the future for more details on the specifics.