Why cybersecurity is not an autonomous vehicle's only vulnerability

Cybersecurity is a major concern for autonomous vehicles, with think tank RAND Corporation identifying four major types of cyber hacking: the hacking of the connected roadway to cause traffic disruption or damage, the theft of vehicle information, the hacking of vehicles to take control of driving, and ransomware attacks that lock people in a vehicle or deactivate a vehicle.

But is cybersecurity a AV’s only vulnerability?

As more AVs have been deployed by companies such a Cruise, Waymo and others, the reports of pranks and physical vandalism have become increasingly common.

From immobilizing a robotaxi by covering its sensors, to taking a hammer to it or setting it on fire, several incidents have occured that are costly to the service providers and damaging to their brand reputation.

Researchers have long noted the ease with which traffic signs can be moved, painted over or manipulated to mislead the guidance systems of AVs.

Anti­car activist groups, such as Safe Street Rebel in San Francisco have published articles and videos on how to disable AVs by covering or damaging their sensors and other on-board equipment.

'Griefing' against AVs

Ghost vehicle studies by Stanford University researchers have revealed how many pedestrians and road users behave aggressively toward AVs at road crossings and junctions.

Verbal abuse, blocking and bullying the AV to achieve a more advantageous position in the traffic flow are occurring more often than many might think.

In fact, such behavior is so common that researchers use the term “griefing” to describe it.

The word captures the sense that the people are doing more than “bullying” the AV physically. The behavior also represents symbolic and communicative expressions of protest against the technology.

So, how should AVs protect themselves and their passengers from abuse?

Some protection is provided by the arrangements adopted at many of the AV depots. Fenced in compounds, manned security gates and extensive CCTV coverage provide protection against misuse or vandalism when the vehicles are recharging or undergoing maintenance.

And one early on-road response, advocated by Volvo, has been to avoid looking or behaving differently from human-driven vehicles.

This school of thought maintains that AVs should look and behave as much like traditional road vehicles as possible to avoid drawing attention.

And another response has been to program the AVs to be more assertive on the road.

Waymo and others have adjusted their driving algorithms to be more assertive at stop signs and junctions, limiting the ability of pedestrians and other road users to take advantage of the AV’s safety-first approach.

But the continuing stream of incidents seems to suggest that such early adjustments have not been sufficient.

AVs appear to be too complex to avoid attempts at interference and manipulation, and also too costly to not have components that are worth stealing.

While few thefts have been reported to date, AVs contain valuable components that will attract ongoing attention.

One way to bypass theft and damage and protect AVs and their occupants may be to use more robust sensors, similar to those found on military vehicles and to use external body panels manufactured from resistant or easily replaceable materials.

It appears unavoidable that specialist software will be required to analyze camera data in real-time to identify griefing.

The combination of on-board situation awareness and human remote monitoring from control rooms will probably need to be better integrated with the fleet of roadway assistance vehicles which bring the humans when an AV gets stuck.

The roadway assistance protocols need to be extended to include the various categories of griefing and the various appropriate responses. And, finally, a degree of integration with private security as well as police systems may prove necessary for dealing with the more extreme forms of griefing.