Uber technology chief Raquel Urtasun is certain of one thing — that there will be self-driving cars, trucks (and possibly boats) in our future, and civilization will be better for it.
She’s not quite as certain about how soon autonomous vehicles (AVs) will be available to the masses, what the global cultural implications might be, or who will be held liable when things go wrong. But then again, that’s not her department.
“It’s my job to make sure [accidents] don’t happen,” she told Reuters Global News Editor Alessandra Galloni during a Reuters Newsmakers event on April 8.
Urtasun is one of the world’s foremost experts on artificial-intelligence applications for AVs. Uber hired her two years ago to head the company’s Advanced Technology Group in Toronto, where she and her team are working to solve one of the most formidable technical challenges of our time: How to introduce millions of driverless vehicles into the chaotic slipstream of daily traffic without killing anyone.
Her mission gained renewed urgency after a pedestrian was struck and killed last year by one of Uber’s self-driving cars in Arizona, calling the company’s technology into question. The success of Uber’s self-driving technology is seen by many analysts as the key to the company’s long-term viability, a lingering concern for investors ahead of the company’s highly anticipated IPO, which could come as soon as this week. Since the Arizona accident, Uber has scaled back its predictions about when autonomous vehicles will be widely available, and has re-dedicated itself to developing better, safer, more cost-effective solutions to the technical puzzle of driverless transportation.
Uber’s long game all along has been to establish dominance in the ride-sharing space, then to replace its global network of independent drivers over time with autonomous vehicles that don’t have to be paid.
“This is a long-term effort, not something we can engineer and deploy right now,” Urtasun explained. “Science, cost, and regulation — these three things need to come together” before AVs are an everyday reality. Urtasun is working on the science.
Specifically, Urtasun and her team are responsible for developing the “brain” of Uber’s self-driving fleet, the complex alchemy of cameras, sensors, GPS, satellite mapping, robotics, communications infrastructure, and artificial intelligence that allows a driverless car to “see” the road, avoid obstacles, and respond to immediate, unpredictable dangers. “If a human brain can do it, there is nothing in principle preventing a robot from doing it,” she said, “but we are not there yet.”
When asked where the technology will be in ten years, Urtasun would only speculate that a combination of self-driving cars, public transportation, and owner-operated vehicles will be available then. She did however paint an attractive picture of how the world will improve once AVs are an everyday reality.
Noting that “1.3 million people die in car accidents every year,” Urtasan said Uber is “developing the technology to prevent that from happening.” Furthermore, in a world of driverless Uber cars, “people will have more choices and freedom.” Children and the elderly will be more mobile.
People won’t have to own a car if they don’t want to. There will be up to 90% fewer cars on the road, she explained, which will be safer and better for the climate. “Fewer cars also means less parking, opening up public spaces” to more parks, bikeways, and pedestrian-friendly areas.
Of course, Uber’s long game all along has been to establish dominance in the ride-sharing space, then to replace its global network of independent drivers over time with autonomous vehicles that don’t have to be paid. But the company’s larger vision is to transform the whole idea of public transportation to include not only driverless vehicles, but also buses, bikes, and even flying taxis. According to Uber CEO Dara Khosrowshahi, the company envisions itself as “global A-to-B urban mobility platform” that will slowly but inevitably nudge aside one of the 20th century’s most iconic inventions, the personally owned and operated automobile.
The role of artificial intelligence
Urtasun also provided an intriguing glimpse into the role artificial intelligence will play in the evolving development of driverless technology. Because infrastructure changes would be too costly and time-consuming, said Urtasun, Uber is developing individual vehicles with an AI system capable of “self-learning” the geography of its immediate environment, the ability to “reason” in real time, and to “act differently,” depending on wherever it is in the world and whatever the driving conditions may be. But improvements in sensors, computing, and AI algorithms are still necessary to perfect this technology, she added, especially when it comes to “scaling” the technology to accommodate millions of vehicles at once.
Urtasun did not downplay the technological challenges involved, and also allowed that open collaboration with other AV developers in other companies would be necessary to accomplish the company’s goals. Interestingly, Urtasun said Uber is much more “open” to these types of collaborations now than it was just a few years ago. “Uber has a vision of where we need to go and understands that being closed is not the best way to get there,” she said.