A Trusted Decision-Making System for Autonomous Driving in Supply Networks with Untrusted Components
Researchers want to develop a trusted decision-making system for autonomous vehicles (AV) to take real-time actions that consider untrusted components in a mixed driving environment involving human-driven vehicles.
Funded by the CCI Hub
Project Investigators
Primary Investigator (PI): Katherine Smith, Old Dominion University’s Virginia Modeling, Analysis & Simulation Center
Rationale and Background
AVs’ current decision-making technologies use information from other AVs to understand the driving environment.
However, a mixed environment will include unpredictable human behavior (driving aggressively, sudden deceleration/acceleration, making sudden lane changes), while the vehicle to everything (V2X) features of AVs make them vulnerable to false information transmissions and cyber attacks.
Current AVs make control decisions in real-time to ensure driving safety when anticipating a possible collision with a nearby vehicle. However, they ignore the effects of emergency actions on surrounding vehicles, which can lead to accidents.
In addition, AVs assume information from V2X communication can be applied without accounting for the possibility that messages are vulnerable to cyberattacks, such as the transmission of false GPS signals.
There is also increased concern about mixed driving environments in ports, as disruptions to their operations can lead to widespread ripple effects on global supply chains.
Methodology
Researchers plan to:
- Collect data for analyzing aggressive driving behaviors and false information transmission.
- Develop and design identification and detection method for a trusted decision-making system.
- Conduct tests and sensitivity analysis on the designed decision-making system.
Simulations and the framework itself will be designed and tested in a digital and cyber physical twin of the Port of Virginia to study the effects and validate the model in the domain of supply chain cybersecurity.
Projected Outcomes
Researchers’ new trusted decision-making system will extend their current research project, which creates a cyber-physical space (CPS) and digital twin of the Port of Virginia.
They envision a framework that will test the driving safety of multiple AVs conducting control decisions under different scenarios in the transportation network connecting the Port of Virginia ecosystem.
This will involve extending the CPS of the port’s scaled representation of the trusted decision-making system and will consider human-driven behavior and false information transmission.