IOE-STAGE SITE
U-M IOE faculty have been awarded second place at the NATO Innovation Challenge for their presented solution that integrates optimization and human factors for autonomous decision making.
U-M Industrial and Operations Engineering (IOE) Assistant Professor Jessie Yang and Associate Professor Cong Shi have been awarded second place among 80 teams from NATO countries and a monetary award for their presented solution at the Fall 2020 NATO Innovation Challenge.
The most recent NATO Innovation Challenge focused on seeking innovative ways to establish and strengthen the trust between humans and autonomous systems. The purpose of the Challenge more broadly is to “tap into the minds” of those within NATO countries. The solutions proposed in the Challenge, if accepted, become essential in ensuring security and accurate response to crises, both within NATO and in other organizations as well.
Yang and Shi were invited to participate in the challenge because of their recent work in this area. “To enable effective teaming, trust is a central factor,” said Yang and Shi. “We are thrilled to know that NATO has recognized the importance of investigating human-autonomy teaming.”
“One individual could be a rational decision-maker whose trust dynamics can be approximated by Bayesian inference. While another individual could be less rational and distrust the autonomous agent regardless of its reliability.”
Jessie Yang,
Assistant Professor, U-M Industrial & Operations Engineering
The solution presented by Yang and Shi is entitled “A Personalized Approach for Predicting and Promoting Trust in Human-Autonomy Teaming.” The central aspect of their solution is that it conceptualizes trust as a dynamic variable that can change over time due to moment-to-moment interactions between the human being and the autonomous system.
In their prior work, Yang and Shi have noted that human agents have different types of trust dynamics, “One individual could be a rational decision-maker whose trust dynamics can be approximated by Bayesian inference,” said Yang. “While another individual could be less rational and distrust the autonomous agent regardless of its reliability.”
The solution involves incorporating estimated real-time trust in a Markov Decision Process with the aim of enabling an autonomous agent to explicitly consider a human agent’s trust when making decisions, and therefore use different strategies when interacting with different types of human agents.
Jessie Yang joined U-M IOE in 2016. Beyond human-agent teaming, Yang’s main research interests include human factors in high risk industries. She is also a core faculty member of the U-M IOE’s Center for Ergonomics and an affiliate faculty member of the U-M Robotics Institute.
Cong Shi joined U-M IOE in 2012. His primary research interests are focused on the design of efficient algorithms with theoretical performance guarantees for stochastic optimization models in operations management.
