Focused on the Road Ahead

Ever wonder what would happen if someone hacked your smart car? Raphael Stern is part of a group of researchers helping prepare cities for automated vehicles and the problems they might bring. Stern’s focus is cybersecurity.

The Center for Connected and Automated Transportation (CCAT) was created by the United States Department of Transportation (USDOT) and the University of Michigan, along with its partners as part of the Fixing America’s Surface Transportation Act. CCAT, based at the University of Michigan, received $15 million dollars to help prepare US cities for connected and automated vehicles. 

As part of the center, Stern is researching cybersecurity. Cybersecurity research in transportation encompasses individual vehicles and transportation systems. As vehicles increasingly rely on technology, there is potential for malicious actors to interfere with automated vehicles. Car manufacturers and other researchers are very interested in vehicle-level safety, some of the most obvious instances of interference might cause a car to stop suddenly or drive off the road. 

Stern’s interests, however, lie at the system-level, hardening our transportation systems against cyberattacks. The interesting problem for Stern is subtle, system-level interruptions of the traffic flow. He is interested in stealth attacks, subtle, difficult to spot anomalies in the way individual vehicles drive that could cause significant interruptions in traffic flow. He researches how to detect such anomalies in traffic flow.

So how does one study something that will be hard to determine and may not even be happening yet?

Stern and graduate researcher Tianyi Li study traffic as a flow problem, with individual vehicles modeled as particles. They look at normal and abnormal variations in traffic flow and try to determine which anomalies could stem from a non-human cause. 

Stern’s earlier research looked at how individual vehicles influence traffic flow and what effect the insertion of a few automated vehicles would have on traffic flow. He also studied how one could identify an automated vehicle in the space-time diagrams that transportation engineers use to study traffic flow. 

Now, Stern and Li propose a novel approach, using artificial intelligence (AI) backwards. AI usually analyzes many instances and then “predicts” a similar instance. Instead, Stern and Li look at one vehicle driving pattern, compare that to a large data set, and determine if this one instance looks like it belongs to the set. They ask, what are the chances that this instance came from that array? 

Stern says, “A cyberattack like this is not imminent, but I do think it is something we want to be prepared for. I do not worry about it when I am out driving. I am focused on the road ahead.”