Supernumerary robots are assistive devices that may attach to the human body and be controlled to perform coordinated actions with a person's natural limbs. Supernumerary robots have potential to reduce risk and increase human capability in physically-demanding and/or dangerous workplace activities, including construction, agricultural applications, search-and-rescue, and outer space exploration. The research objective of this project is to advance a fundamental understanding of human-machine interaction within the context of real-time dexterous control of a "third arm" supernumerary robot. The project will develop a novel human-machine interface that will allow a human to use electromyographic activity (EMG) to command the robot in a closed-loop manner using multimodal sensory feedback derived from visual and supplemental vibrotactile information. The project team will use the new system in two sets of human subject experiments designed to explore the bi-directional adaptation that occurs within the human-robot dyad as the human learns to control the machine in a novel three-handed manipulation task, even as the robotic controller adapts to the changing signal properties of muscle EMGs as the human user fatigues. The project advances the National Science Foundation's mission to promote the progress of science, to advance national health, prosperity, and welfare, by advancing a fundamental understanding of co-adaptive human-machine partnerships. Broader impacts of this project include establishing a summer program for elementary and middle school female students to introduce principles of robotics, sensors, neuroscience, and human performance. The summer program is directly aligned with the National Science Foundation's goal to support activities designed to increase the participation of women and other underrepresented groups in science and technology.
This project explores human-machine cooperation within the context of a "third arm" supernumerary robot. Two sets of human subject experiments are planned. The first establishes the feasibility of a novel, adaptive, EMG-based methodology for haptic-guided closed-loop control of a supernumerary robotic arm's end-effector. Several biosignals sensitive to cognitive state will be recorded, including pupillometry, galvanic skin response, heart rate variability, and electroencephalography (EEG). Subjects will perform a point-to-point trajectory task and a 3D version of the Fitts's Law task to assess control efficiency with and without vibrotactile performance feedback applied to the user's foot. The second explores the coupled dynamics of short-term human learning of a three-handed manipulation task that requires the user to coordinate motions of their two hands with those of the supernumerary robot to manipulate a novel three-element object to recreate specific desired poses. At the same time, online computational learning adaptations of the robot's grip force will be driven by perturbations of object kinematics and kinetics caused by the user's natural arms. The work will advance a fundamental understanding of (1) human interface performance in physical workspaces where cognition is embodied in both the human and supernumerary robot; (2) cognitive load and neuroplasticity using EEG and other physiological measurements; and (3) improvements in tri-manual coordination and manipulation capability arising from short-term training.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
