Don't Give Up on Self-Driving Cars Now
Two weeks ago, one of Uber’s self-driving cars tragically killed a 49-year-old pedestrian in Tempe, Arizona. This was the first fatality from an autonomous vehicle — which is quite remarkable considering that Uber’s cars have driven approximately 2 million miles. Last week, a Tesla Model S had a fatality during autopilot in California, after 5 billion miles. Tesla’s cars, which feature self-driving capabilities, though are meant to be operated with a human driver always at the wheel, are somewhat different to fully autonomous vehicles, with the crashed car reporting that the human driver did not have his hands on the wheel despite alerts prior to the crash.
Current driving data on fully autonomous vehicles is nowhere near the level of significance required to adequately compare their safety to those piloted by humans. While there were nearly 6,000 pedestrian fatalities in 2016 — or one every 1.6 hours — self-driving cars have driven in total approximately 1/500 of what Americans drive every year. The long-term opportunities autonomous vehicles create for improving safety should not be shelved as a result of tragedies that occur during a shorter-term development process.
Uber has shown that it understands the gravity of the situation by halting all testing. Other manufacturers, such as Toyota, have followed suit. Uber has also shown full willingness to cooperate with law enforcement to determine what went wrong, and to remedy the situation as best as possible. It is not possible to make up for a loss of life. But it would be wrongheaded to abandon the advancements that promise to one day save those thousands of lives lost every year to human-caused traffic accidents.
Preliminary findings by the Tempe Police Department have shown that Uber’s technology bears no fault in the crash. This accident, like many other pedestrian collisions, occurred under conditions of low light and away from a crosswalk. According to Tempe Police Chief Sylvia Moir, “it’s very clear it would have been difficult to avoid this collision in any kind of mode [autonomous or human-driven] based on how [the victim] came from the shadows right into the roadway.”
This suggests that it would be helpful to have more, not fewer autonomous vehicles in circulation. Why? Increasing the amount of connected technologies on the road would enhance the communications capabilities that would allow autonomous vehicles to identify risky situations and respond even faster. This risk reduction could be facilitated by greater involvement from local and state governments, who can use their hard-won understanding of the causes of motor vehicle fatalities to guide the way in which roads are constructed. This government buy-in would make transit via self-driving car safer.
Human drivers have been getting more dangerous because of texting and driving and other distractions that have lowered driver responsiveness. Ninety-four percent of all accidents are related to human error, with the Uber fatality falling into that 6 percent of unavoidable accidents given what we know from early reports by Tempe PD.
Extreme responses — such as that of Paris, where all autonomous vehicles have been banned within city limits — should be avoided. Such responses will reduce the long-term potential for safer, more livable cities that autonomous vehicles promise. Another option that is gaining popularity, limiting ownership to shared fleets, is similarly a losing proposition for safety and urban design. This latter option is what has been proposed by both Uber and Lyft via their Shared Mobility Principles for Livable Cities initiative.
Self-driving cars have the potential to revolutionize and improve the lives of urban dwellers. They have no need to be idle, as human driven cars do 95 percent of the time. If privately owned and rented out through platforms like Uber — rather than restricted to ownership by shared fleets — they could not only reduce the demand for car ownership but also serve as passive income for those who do own cars. This incentive would increase efficiency, thereby lowering congestion and cut back on the need for the staggering 800 million parking spaces in the U.S. This would, in turn, free up land for more productive uses such as businesses, houses, or recreational areas. Urban design could be refocused on building safer and more pleasant environments.
The opportunities afforded by automated driving are not limited to making commutes easier. Self-driving cars have profound implications for road safety, urban design, and the cost of living. Of course, they are still in early stages, and need input from communities and regulators to achieve a tolerable level of safety. The recent tragedy in Tempe should not prevent these innovations from ultimately improving and saving lives in the future.
Ryan Khurana is a tech policy analyst in Washington, D.C. He is a Young Voices Advocate.