Human-AI Interaction Design for Autonomous Vehicles During Security Events

Objectives

  •  Understand  users’ perception of and response behavior to cyberattacks in autonomous vehicles 
  •  Derive  design requirements for user response to such security threats
  •  Identify  factors an AI system should be aware of regarding driver behavior to support driver response

Method

Results

Phase 1

  •  Adapt  interface design based on attack severity and phase of driver response. 
  •  Prevent  information flooding during the initial stages of an attack. 
  •  Enable  vehicle control through internal modules and physical components and firewall the vehicle from receiving external connections

 

Phase 2

  • Autonomous systems be designed to  communicate  risk and  gauge  the current level of user trust
  • AI designs could emphasize the system’s capacity to take actions aimed at gently rebuilding trust when it has been eroded, taking into account the human occupant’s behavioral responses

Publications in Human Factors Journal (2024), HFES 2023, and VehicleSec 2024

 

Lim, C., & Rajivan, P. (2023, September). Who hacked my car? Designing autonomous vehicles to support driver response to security threats. In Proceedings of the Human Factors and Ergonomics Society Annual Meeting (Vol. 67, No. 1, pp. 247-252). Sage CA: Los Angeles, CA: SAGE Publications.

Lim, C., Prendez, D., Boyle, L. N., & Rajivan, P. (2024). The Impact of Cybersecurity Attacks on Human Trust in Autonomous Vehicle Operations. Human Factors, 00187208241283321.

Lim, C., Xu, T., & P. Rajivan (2024). Is My Car Hacked? Analyzing How Security Attacks During Autonomous Vehicle Operation Impact Trust in AVs. In Symposium on Vehicles Security and Privacy (VehicleSec) 2024.

A) Phase 2. Experiment Procedure

B) Phase 2. Attack Drive Snippet (you can also watch it here)