One of Canada’s leading artificial intelligence researchers is launching a new autonomous-vehicle startup with major backing from U.S. investors—including her former employer, Uber.

Founded by Raquel Urtasun, Waabi launched Tuesday with an US$83.5-million Series A round led by Khosla Ventures, with participation from Uber, OMERS Ventures, the Business Development Bank of Canada and Aurora Innovation, among others. The round also included Sanja Fidler, Urtasun’s U of T contemporary and Nvidia’s AI director; and British Canadian computer scientist Geoffrey Hinton, considered the godfather of deep learning technology; and Radical Ventures, whose managing partner Jordan Jacobs was, with Urtasun, Hinton and others, a co-founder of the Vector Institute for Artificial Intelligence. The startup claims the Series A is among the largest ever raised in Canada.

Urtasun, who will be Waabi’s CEO, is one of Canada’s highest-profile AI researchers. A professor in the University of Toronto’s computer-science department, she was the Canada Research Chair in Machine Learning and Computer Vision before joining Uber in 2017 to lead a Toronto outpost of its Advanced Technologies Group as chief scientist and head of research and development. 

Waabi’s launch reveals what Urtasun has been working on since the dissolution of Uber’s Toronto R&D team, which did not join Aurora Innovation when the ride-hail company sold its self-driving division in December 2020. Waabi’s backers, which include both Uber and Aurora, suggest Urtasun’s influence hasn’t waned—and that Toronto remains firmly in the technological race to make autonomous vehicles. 

Waabi’s team of 40 and counting is split between Toronto and California, though the company said the majority of the team will be in its Toronto headquarters.

“The engineering team is headquartered in Canada; we plan to definitely always have a very big presence in Canada,” Urtasun told The Logic. The company’s IP will also reflect the Canadian headquarters and not be offshored, she said, adding that Ontario is also a fairly open jurisdiction in terms of autonomous-vehicle regulation. “There is such incredible talent in this area from the point of view of AI technology,” she said.

Waabi will focus on long-haul trucking, and Urtasun said the company is aiming to augment labour shortages, improve safety and roll out the vehicles in the more predictable highway environment.

“There’s really two reasons for this choice. One is the logistics: there is an incredible need for drivers, right? There’s a big shortage,” said Urtasun. “This also is a domain where—although it is still really difficult—it’s simpler than if you want to do robotaxis in cities like Toronto, where there’s a lot of things that can happen.” 

The technology industry has long faced questions about what will happen to workers like truck drivers if their jobs are automated. But it is a profession that takes a toll on drivers, and has especially during the COVID-19 pandemic, which could push more workers into retirement and make recruitment difficult in what some say is already a short-handed industry.

Waabi is far from the only company making the case for automating heavy-duty trucks. While Uber abandoned its embattled autonomous-truck project in 2018, global players, like Waymo’s Via, and Canadian ventures, like NuPort Robotics and the Alberta Motor Transport Association, are all testing ways to automate more miles on the open road.

Even Waabi backer Aurora is working on its own truck project. The U.S. company’s Canadian co-founder, Chris Urmson, has gone on record with one explanation of the tech industry’s move into trucks: unlike passengers in robotaxis, a “roll of toilet paper” doesn’t care whether a truck’s every twist and turn is perfectly smooth.

Urtasun said Waabi has something other companies don’t: a way to cut down the many, many miles most test vehicles need to drive to train autonomous algorithms for different conditions. 

“After so many years in the industry, we need a radically new approach to be able to really solve this at scale,” said Urtasun. 

Urtasun has framed the venture as the “culmination of my life’s work.” She acknowledged that self-driving cars have gained limited scale despite two decades of progress, in part because of what she called the “manual tuning” of extensive road testing.

“We can train the system—we can test, without such a big requirement of driving and driving in an environment. This results in software that is safer and more affordable,” she said of Waabi’s approach. “It’s better equipped than existing approaches to potentially arrive at the point where we can actually, one day, drive in really adverse weather conditions.”

This technology will help Waabi target Level 4 autonomy, one of the highest levels of self-driving capability. Urtasun said her team is focused mainly on building out its artificial intelligence software and may be open to partnering with hardware makers and vehicle manufacturers in the future.

Urtasun admits the autonomous-vehicle industry has its version of the Trolley Problem, an ethics thought experiment asking where one should steer a vehicle when every option would result in at least one likely death. What happens when it’s AI—not a human—making that call? 

“How do we understand why the system decides to do one thing versus something else? One of the differences of our technology is that it’s actually interpretable … not a black box,” she said. 

Michelle Scarborough, managing partner of strategic investments at BDC Capital, acknowledges that Level 4 autonomy, which is designed to need almost no driver intervention in most conditions, is not something that will come to market tomorrow, even with a self-learning system speeding up the testing process. But she said her team is prepared to invest in multiple rounds for Waabi going forward, calling Waabi’s plan “very viable” and “fully scalable.” Scarborough declined to specify BDC Capital’s investment in the Series A, but said it was a “meaningful” amount.

“I think the differentiator here, specifically related to the competition, is the fact that [Urtasun is] trying to build a fully integrated, very efficiently run system—as opposed to the current engineering design in the market right now,” Scarborough told The Logic.

“She is an incredibly ambitious woman and an incredibly ambitious leader, not just by virtue of her own actions, but by virtue of the people that she’s been able to surround herself with,” said Scarborough, who is also the managing director of BDC Capital’s Women in Technology Venture Fund. 

“When we did all of our calls to with other investors that are participating in this round, and just other people that have worked with [Urtasun], all of them said that she is very open to working with others, to having empathy around her team and being able to continue to see the big picture while she is operationalizing her business.”