In theory, driverless cars could make Toronto’s roads safer, but, if so, when?
As Toronto pedestrian deaths rise, it is hard to know just how auto technology can make citizens safer.

TheStar.com
April 6, 2018
David Rider

As Toronto pedestrian deaths rise , self-driving cars hold the promise of reducing the carnage.

But that dream dimmed last month when a Volvo SUV driving in autonomous mode in Tempe, Ariz., hit and killed 49-year-old Elaine Herzberg as she pushed her bike across a roadway on a Sunday night.

The autonomous driving vehicle was part of ride-hail giant Uber’s multi-city test project.

Uber immediately halted tests, including those in Toronto, where Volvos with bulging sensor-filled roofs have become a familiar sight on roads near U of T’s downtown campus.

That first known case of a self-driving car killing a pedestrian was followed by a Tesla SUV in “autopilot” mode slamming into a California roadside barrier and catching fire, killing the driver.

David Ticoll, author of a 2015 study on the potential impact of driverless cars on Toronto, said the autonomous vehicle (AV) industry has been urging governments to minimize regulatory obstacles to testing and commercial deployment, “so we can save all these lives.

“So obviously this line of reasoning has lost its force as a result of these two events,” said Ticoll, a Distinguished Fellow at the Innovation Policy Lab at the Munk School of Global Affairs.

“Yes, we do want to get these vehicles on the road soon, but it doesn’t mean we throw caution in the wind to do that. That’s not an argument against self-driving cars; that’s an argument for doing things carefully and properly and with due diligence.”

Ticoll and other experts interviewed by the Star have no doubt AVs will eventually make streets safer in Toronto and other cities because computers are more vigilant and quicker to act than humans, who are prone to distraction, fatigue and frustration.

“At the end of the day, I don’t have any doubt that automation will reduce the total number of crashes. I think, more importantly, it will reduce the (severity) of crashes … ,” says James Sayer, director of the University of Michigan Transportation Research Institute.

“I’m of the opinion, without knowing all of the details of the Uber system, that (the Tempe) crash was completely preventable with the technologies aboard. It was a matter of the implementation.”

Most AVs employ GPS, cameras, inertial navigation systems and sensors including LIDAR (light detection and ranging) with 360-degree view and depth perception to identify potential obstacles. Working together, they predict what those obstacles will do and brake, steer or accelerate accordingly.

While the public may expect a revolutionary jump in car design, the experts expect an evolution from the driver assistance systems available, if not yet widely adopted, today.

They include automatic emergency braking, including pedestrian detection, lane-departure warnings and lane assistance to steer the car back between the lines, and adaptive cruise control that adjusts speed automatically to keep an appropriate distance from traffic ahead.

Cars are getting smarter, but 14 pedestrians died on Toronto streets so far this year, compared with nine at the same point in 2017?

Last year, saw the worst such death toll in more than a decade, as motorists killed 46 pedestrians and cyclists on the city’s streets.

Canadian and U.S. transportation agencies are unable to point to any data suggesting so-called semi-autonomous technology has reduced the number of serious injuries or road deaths.

Part of the problem is that the safer future most experts see has few or no cars driven by people. We are in the earliest stage of intermingling of driverless cars and cars piloted by drivers. It’s analagous to the early 1900s, when automobiles started rumbling alongside horse-drawn buggies.

Cyclists and pedestrians will always be with us.

“One of the biggest challenges that AVs have is to make them capable of handling the random behaviour of human-operated vehicles and other items in their path like pedestrians or cyclists,” says Ross McKenzie, managing director of the University of Waterloo’s Centre for Automotive Research.

“If we had a dedicated roadway that only had AVs on it, you’d probably have some AVs now, because the technology is there,” says McKenzie. His centre has four self-driving cars which are close to completing their testing on a private track and the vehicles will soon venture out on public roads.

Real-road testing is done under the Ontario transportation ministry’s 10-year pilot project, which requires the provincial approval of participants, that a human driver always be ready to take over and that vehicles be equipped with a prominent “off” button to disable autonomous features instantly.

Ontario has approved seven participants, for a total of 10 AVs: Uber, Magna, The University of Waterloo; The Erwin Hymer Group; QNX, Continental, and X-Matik Inc.

No participant has reported injuries from a crash, but most testing has been on closed courses, something that is expected to change over the life of the pilot, which ends in 2026.

If the “Sidewalk Toronto” project proposed for the Quayside district on Toronto’s east downtown waterfront proceeds, the city could have a novel testing opportunity.

“We envision Quayside as a district designed for autonomous vehicles as well as transit, cyclists and pedestrians, to the exclusion of the traditional automobile,” says Rohit Aggarwala, head of urban systems for New York-based Sidewalk Labs, which won a Waterfront Toronto competition to spend a year planning high-tech test solutions for Quayside and potentially beyond.

He envisions residents sharing autonomous “taxibots” or bigger “vanbots” for trips, where walking, biking or transit don’t make sense, either because of the destination or the needs of passengers.

“The biggest opportunity at Quayside is less ‘how do you make the vehicle a smart driver,’ (but) really … more about ‘how does this new kind of vehicle fit into an urban environment with a dynamic curb pickup and drop-off as opposed to a (parking) storage area for unused vehicles …’,” Aggarwala said.

“In a world where you can always count on the vehicle to follow the rules, that simplifies the environment and that should lead to greater safety for all concerned.”