TRUST BUILDING IN HUMAN-AUTONOMY TEAMING: A REINFORCEMENT LEARNING APPROACH
Abstract
As autonomous and robotic systems become more capable in perception, planning, learning and action, there is an increasing possibility that they will become full-fedged team members. The humans and autonomous agents are expected to work as a team in environments subject to uncertainty and dynamic changes. To enable effective teaming, trust has been identifed as one central factor. To build trust in human-autonomy teaming, this proposal aims to to develop algorithms that enable the autonomous agent to infer the human s objectives and moment-to-moment trust and to use different interaction strategies for building trust and enhancing team performance. We use a transformative approach by combining theory-driven human factors models and data-driven computational methods. In the proposed effort, we consider human-autonomy teaming a in highworkload time-critical intelligence, surveillance and reconnaissance (ISR) mission. We will develop a simulation testbed and conduct three human-in-the-loop experiments. Data collected in the experiments will be used to develop and validate the proposed algorithms. The anticipated outcomes include a full-scale dynamic-trust-driven computational model for optimal autonomous decision making, and a game-theoretic model for human-autonomy teaming.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2021
- Source ID
- FA95502010406
Entities
People
- Xi Yang
Organizations
- Air Force Office of Scientific Research
- Board of Regents of the University of Michigan
- United States Air Force