An experimental autonomous vehicle arrives for a drop-off on Dec. 8, 2020, in Miami Beach, Fla.Carl Juste/The Associated Press
Who among us will ever forget the “Conversation Boom” of July 8, 2022? For almost an entire day, a nationwide Rogers network outage left millions of Canadians with nothing to do but take long undisturbed walks and talk to each other. Debit services were unavailable online or at checkouts, and e-transfer was down. The Toronto Police Service reported problems for some people calling 911.
I was hard at work in a café with other wifi nomads working on a story about bicycle lanes when my editor suggested I pivot to explore how susceptible the fully autonomous vehicles of the future may be to an outage like this.
“It was not a great day for AVs (autonomous vehicles), to have this critical link go down,” said Steven Waslander, a professor at the University of Toronto Institute for Aerospace Studies and the director of its Toronto Robotics and AI Laboratory. “[AVs] are mostly expected to be independent of networks for exactly this reason, but can benefit immensely from 5G throughout and external control from remote pilots, particularly in development, when cars pull over if [they are] confused or stuck.”
As the cyber-smoke cleared and service was restored, I interviewed Amir Khajepour, a professor at the University of Waterloo in the Department of Mechanical and Mechatronics Engineering and a Canada research chair in Mechatronic Vehicle Systems. He is the lead researcher on the WATonoBus project, which started in August, 2021. The self-driving bus navigates a 2.7-kilometre loop that features traffic lights, delivery trucks, pedestrians and even meandering geese at a top speed of 20 kilometres per hour.
He put the Rogers network outage and AVs into context explaining that the car would be able to operate to safely stop and definitely wouldn’t endanger passengers. He said the car would require connectivity to operate normally, but “there will be enough redundancy in system to ensure connectivity all the time.”
He said vehicles always have redundancy for safety-critical systems, which could come in the form of subscribing to more than one network or a satellite connection.
How would an outage like the one we just experienced in Canada affect ADAS (advanced driver assistance systems, which are features many cars today have including lane-keep assist and adaptive cruise control) and AVs (autonomous vehicles are ones that won’t require a steering wheel)?
For any safety-critical application including ADAS and autonomous driving, there are many layers of monitoring and safety to handle cases such as loss of connectivity. In autonomous driving and some ADAS, access to real time data and Vehicle-to-Everything (V2X) are critical, meaning that in case of loss of connectivity, the vehicle will hand over the control to the driver, or in case of higher level of autonomy, it has enough on-board sensors and information of the driving environment to operate in safe mode to pull over and stop.
Is it true that an outage like the Rogers outage would not jeopardize vehicle safety? Autonomous vehicles could still pull over, stop etc. It would cause problems with connectivity (no GPS, no updates).
This is absolutely correct. All these scenarios are considered at the design phase of any ADAS and automated driving to ensure they have enough redundancy to safely hand over the operation to the driver or pull over and stop in case of level 3 (where a driver must be present and awake, but can watch a movie) and higher.
Would it have a worse effect on city streets than on highways?
If our discussions are limited to ADAS and automated driving at the present time, there won’t be any effect. If we are expanding this to general traffic control and smart cities, probably the main effect on city streets will be on real-time control of traffic lights.
Anything else you might think worth addressing?
I think the perception of most people from autonomous driving is a car that can drive like a human in any situation and condition. This is the cause of many misconceptions about autonomous driving. I think helping to inform the public would be very helpful and important.
Why do you think this misconception has taken hold? Is the technology too complex? Has the media been lax in reporting? Are we just drawn to the most dramatic scenario when it comes to transportation?
I think for any new technology, and not just autonomous driving, we tend to be drawn to the ultimate scenario that may not be even possible or materialized technologically or economically. Lack of enough public experience in using any new technology will keep this misconception going for a while. I am sure if you ask anyone who has experience driving a vehicle with some automated driving features, they will have a completely different view and would never interpret autonomous driving as something even remotely close to a human driving. This misconception will go away eventually, but more public awareness, I think, would be very helpful.
This interview has been condensed and edited.