Anyone who books an Uber in Pittsburgh in the coming weeks may discover that the person behind the wheel is also a passenger.
Uber will offer customers rides in robotic taxis within a matter of weeks or days. The company has been developing the technology for the past year and has been testing it on the streets of Pittsburgh. It will launch with about a dozen taxis, with the expectation of having 100 on the road by the end of the year. The taxis will have drivers who can take control in an emergency.
Pittsburgh was chosen as the location for Uber’s automated driving project because of its proximity to the renowned robotics research hub at CMU. The city also offers challenging roads and environmental conditions.
Uber is rushing ahead with the technology because automated driving promises to upend the taxi industry by doing away with the need for human drivers. The company has lost billions of dollars in recent years offering incentives to drivers in order to increase its market share. However, the technology is still nascent, and it remains unclear how quickly drivers could be removed from the equation.
The trial also highlights the remarkable speed with which the technology is being rushed to market, and it could play a key role in shaping both the public perception and government regulation of self-driving cars. This is especially true after a fatal accident in June involving a partially automated Tesla. The National Highway Traffic Safety Administration is currently investigating that accident (see “Tesla Crash Will Shape the Future of Automated Cars”). It will also be interesting to see how passengers and drivers react to the prospect of machines taking over what had become an easily accessible new job.
First and foremost, though, Uber’s experiment will most likely generate a great deal of buzz, since it will offer many people their first taste of automated driving. For all the attention that self-driving cars get, few people have experienced the uneasy, awe-inspiring feeling of taking their hands off the wheel (or watching someone else do it), and then having their car drive itself.
There is a dash within the automotive industry not to be left behind as computerization and automation threaten to upend many aspects of transportation. Automated driving systems are now being developed at a remarkable pace, not only by traditional automotive manufacturers but also by newcomers emboldened by the opportunities enabled by the use of sensors, computers, and software.
Indeed, the advent of automated driving now appears inevitable. “The technology is foregone, and the markets are foregone,” says Red Whittaker, a professor at Carnegie Mellon University who did pioneering work on outdoor robot navigation in the 1980s and 1990s, and who led the team that won DARPA’s first automated driving contest in 2007. “It's something that's going to be evolving for decades, but it's going to become part of how things are done.”
Automated driving has been under development in commercial and academic research labs for decades. Over the past few years, however, it moved toward commercialization with incredible speed. Most visibly, Google has tested its self-driving cars over millions of miles of road around Mountain View, California, as well as in Austin, Texas, and Phoenix. Over the past few years, most carmakers, as well as a handful of startups, have begun testing their own automated vehicles.
Ragunathan Rajkumar, a professor at CMU who is collaborating with General Motors on automated vehicle technology, says the Pittsburgh experiment will raise public awareness about how driverless systems work. “Most people of course will be thrilled to be part of the process,” he says. But he adds that Pittsburgh will be a challenging setting, given its weather and the complexity of its roads. He also notes that automated cars are still incapable of dealing with the unexpected. “This is an early testing version,” he says. “It's going to be a long time before you take the driver out of the equation.”
Uber, which began a ride-sharing service in 2009, bolstered its self-driving project, somewhat controversially, last year by hiring dozens of robotics researchers away from CMU. The company has invested hundreds of millions of dollars in the effort, and has promised to spend more than a billion in coming years. It also recently acquired a startup developing self-driving trucks (see “Uber Is Betting We’ll See Driverless 18-Wheelers Before Taxis”).
There are a number of obstacles, including safety issues. While accidents such as the one involving Tesla’s Autopilot feature could turn public opinion against self-driving vehicles, some experts believe that the technology will prove too beneficial to resist. “Yes, there was an accident,” says Raj Reddy, a professor at CMU. “But compare that to the number of fatal accidents there are every day. One accident doesn't crash the whole industry.”
Reddy, the founding director of CMU’s robotics institute in 1979, says government policy and legal liability are the other key obstacles. “But I think all of those things will take care of themselves,” he says.
Uber is not the first to test the idea of self-driving taxis, either. An MIT spin-off called nuTonomy is already running a limited service on the streets of Singapore. But Rajkumar cautions that both the Singapore and Pittsburgh trials may highlight the remaining challenges for automated vehicles. “I think people should temper their expectations,” he says.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Unpacking the hype around OpenAI’s rumored new Q* model
If OpenAI's new model can solve grade-school math, it could pave the way for more powerful systems.
Generative AI deployment: Strategies for smooth scaling
Our global poll examines key decision points for putting AI to use in the enterprise.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.