Uber opening Toronto research hub for driverless car technology

A University of Toronto computer science professor will lead Uber's new research hub devoted to driverless car technology. Uber was drawn to Toronto because the city has been "at the forefront" of artificial intelligence research.

Thestar.com
May 8, 2017
By Kate Allen

Uber is launching a research group devoted to driverless car technology in Toronto, creating a third hub - its first outside the U.S. - for the company’s ambitions in a frenzied field that Uber and its competitors believe will upend transportation, generating billions of dollars in the process.

The Advanced Technologies Group will be led by Raquel Urtasun, a University of Toronto computer science professor who holds a Canada Research Chair in machine learning and computer vision. Urtasun uses artificial intelligence, particularly deep learning, to make vehicles and other machines perceive the world around them more accurately and efficiently.

The group will hire “dozens” of researchers and engineers in the next few years, the company says. Uber will also make a multimillion-dollar, multi-year commitment to the Vector Institute, the artificial intelligence institute that launched in Toronto in March with investment form both government and private sources, including technology companies as Google, Nvidia and Shopify. Urtasun, who will remain a professor at U of T, is one of the Vector Institute’s founding members.

“Toronto has emerged as an important hub of artificial intelligence research, which is critical to the future of transportation,” Uber CEO Travis Kalanick wrote in a blog post welcoming Urtasun and announcing the group, praising the Ontario and federal governments for their investments in the field.

Uber has already put semi-autonomous cars on the roads in Pittsburgh and San Francisco, the homes of the other two Advanced Technologies Groups, and more recently in Arizona, but the company has not said when Torontonians might be able to hail a self-driving car.

Last year, Ontario became the first province to launch a pilot program that permits on-road testing of automated vehicles (AVs). Carmakers and technology companies that want to participate must apply to the Ministry of Transportation, and if approved must have a human in the driver’s seat of the AV at all times.

Monday’s announcement comes at a critical time for Uber’s self-driving car program. Waymo, Google’s self-driving car unit, launched a lawsuit against Uber in February, accusing the company of using stolen trade secrets to develop sensors for its autonomous vehicles.

Waymo alleges that a former manager, Anthony Levandowski, illegally downloaded 14,000 documents, including detailed circuit board designs for a laser-based sensor known as LIDAR, before leaving the company and founding Otto, a startup that was acquired by Uber approximately six months later for $680 million (U.S.).

Uber has denied the accusations, saying that the LIDAR it developed in-house is significantly different from Waymo’s, that after diligent searching there is no evidence of any stolen Google documents on Uber’s servers, and suggesting that the Waymo lawsuit is an attempt to stifle the progress of a major rival.

Waymo has asked a federal court in San Francisco to halt Uber’s self-driving car research until the case can be resolved. The judge is expected to rule on the preliminary injunction within days.

Lawsuits of this kind are common with rapidly evolving technologies for which even a nanometer-thin competitive edge can offer a critical advantage, legal experts say.

“You have people racing to get there with limited (numbers) of people who are well-versed in the field,” said James Pooley, senior counsel at Orrick and a Silicon Valley-based specialist in trade secrets and technology. “People moving from one place to another in a highly charged, highly competitive environment, produces - frequently - cases like this.”

Pooley and others say it would be highly unusual for the judge to order Uber to cease all work on self-driving cars. But even a lesser order could be an impediment.

“They certainly have put an enormous amount of effort into defending themselves. They see this as something that’s quite critical.”

Kalanick has emphasized that self-driving cars are at the core of his vision for Uber and for the future of transportation. “This is not a side project. This is existential for us,” he told the New York Times in an interview after the company acquired Otto.

Urtasun’s lab has focused on making the technology that undergirds self-driving cars more affordable and efficient. She has developed systems that use cameras to reconstruct and understand a vehicle’s surroundings, a much cheaper alternative to LIDAR. She has also worked to find low-cost alternatives to the highly detailed, labour-intensive maps that AVs currently rely on.

Urtasun, who this year was awarded one of Canada’s most prestigious scientific awards - the E.W.R. Steacie Memorial Fellowship from the Natural Sciences and Engineering Research Council of Canada - works on the algorithms that knit all of a vehicle’s sensors together rather than on the hardware itself.

“Toronto and Canada for the past two decades has been at the forefront of AI, and that’s the expertise we’re bringing to Uber,” she said.

A number of self-driving car crashes and mishaps have dominated media attention in recent months. In May 2016, the driver of a Tesla operating in “autopilot” mode died after his car struck a truck turning across his lane. An Uber car in self-driving mode was involved in an accident in Tempe, Arizona, in March.

Police in Arizona cited the other driver in the Uber incident, and a U.S. National Highway Traffic Safety Administration investigation into the Tesla crash blamed human error, noting that Tesla’s crash rate dropped 40 per cent after the installation of autonomous technology.

Improved safety is one of the primary reasons that researchers and regulators are interested in developing self-driving vehicles: algorithms are not prone to the same distractions and mistakes as human drivers, and Ontario’s Ministry of Transportation specifically cites “minimized driver error” as a potential benefit of the technology. The province also cites reduced traffic congestion and greenhouse gas emission as a result of more efficient vehicles.