EAGER: TaskDCL: Mutual Learning and Adaptation for Human-Robot Sensorimotor Interactions in Urgent and Safety-Critical Tasks

This EArly-concept Grant for Exploratory Research (EAGER) project will support research that intends to create a new modeling and learning framework to enable a human and robot to coordinate efficiently through sensorimotor interactions in urgent and safety-critical tasks. Human-robot sensorimotor interactions have shifted the paradigm in manufacturing, transportation, and healthcare over the past decade. However, most existing robots can only function in human-led, highly repetitive, and slow tasks since there is a lack of understanding of human decisions in urgent and dynamic tasks. This project integrates ideas from behavior economics, game theory, optimal control, and human factors to model human decisions under urgency and uncertainty so the robot can leverage such models for safe and fast co-adaptation with the human, with applications in a balancing task when a human rides a robotic bicycle. This research project will promote the progress of science and advance national health by enabling a human and robot to effectively coordinate in urgent and risky situations, which contribute to many failure cases of existing human-robot systems. Intended outcomes from this project will not only improve safety and productivity of future human-robot teams, but also contribute to improved and calibrated human trust to new robotic technologies. The impact of this project will be broadened by new curriculum and undergraduate research opportunities on human-robot sensorimotor interaction, and by establishing an interdisciplinary research community on cycling safety, biomechanics, and assistive technologies. This research project will develop a mathematical framework for human-robot mutual learning and adaptation by studying a novel balancing task jointly performed by a human rider and robotic bicycle. Specific objectives of this project are to 1) model human responses to balancing perturbations when riding a bicycle, 2) develop a game-theoretic robot controller to safely co-adapt with the human rider, 3) understand how the task performance varies with different task conditions originated from the human, robot, and environment, and 4) understand how humans develop reliance on the robot assistance and how that affects their riding behavior in the long term. This EAGER award has been co-funded by the Dynamics, Controls, and System Diagnostics and the Mind, Machine, and Motor Nexus Programs. 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.