Modeling Bi-Directional Trust in Semi-Autonomy for Improved System Performance

Principal Investigator: Dawn Tilbury, University of Michigan,
Lionel Robert, University of Michigan,
Faculty: Xi Jessie Yang, University of Michigan,
Student: Luke Peterson, University of Michigan
Government: Victor Paul, Ben Haynes, U.S. Army TARDEC
Industry: Mitch Rohde, Quantum Signal

This project began in 2017.

Schematic of mutual trust modelIn this project we will examine factors that impact a driver’s trust in the vehicle’s autonomy and vice versa before and after interactions. When one agent meets the expectations of another agent, the willingness to give control over to that agent may increase. Similarly, when an agent fails to meet the expectations, the willingness to give control over to that agent should decrease.

Autonomous and semi-autonomous vehicles have the potential to help drivers successfully and safely complete many military missions while providing the drivers with the flexibility to address other pressing issues. Unfortunately, drivers have failed to fully leverage a vehicle’s autonomy because of a lack of trust in the vehicle’s autonomy. Within the framework of autonomy, we will consider a suite of semi-autonomous behaviors, including maintaining a given speed, staying within a lane (or on the road) as appropriate, avoiding obstacles, maintaining a given distance behind a lead vehicle.

Our goal is to develop methods to predict (1) when the human is likely to take or give control of the driving to the vehicle’s autonomy, and (2) when the vehicle’s autonomy should take or give control of the driving to the driver. To accomplish this, we have the following objectives.

  • To understand and model the set of expectations a driver relies on to determine trust in an autonomous vehicle.
  • Develop a set of expectations the autonomous vehicle should have to determine its level of trust in the driver.
  • To understand how a driver’s individual attributions are likely to impact how much they trust their vehicle’s autonomy and how their vehicle’s autonomy should trust them.
  • To understand how much risk relative to trust is needed in a given situation to trigger either the driver or autonomous vehicle to take control.
  • Develop models of the interplay between the driver’s expectations vs the actions of the autonomous vehicle and vice versa in a given situation to understand the development, decline and repair of trust between the driver and the autonomous vehicle.


  • Petersen, L., Robert, L., Yang, X., Tilbury, D. (2017). Effects of Augmented Situational Awareness on Driver Trust in Semi-Autonomous Vehicle Operation. In 9th Annual Ground Vehicles and Systems Engineering & Technology Symposium (GVSETS). Novi, MI. August 2017.
  • Petersen, L., Robert, L., Yang, X., Tilbury, D. (2018). Enhancing Situational Awareness to Improve Driver Trust,” Submitted to SAE World Conference 2018. Detroit, MI. April 2018.