Autonomous vehicles, both cars and trucks, are being tested in at least a dozen states across the U.S. Commercial use has already begun with autonomous tractor-trailers hauling cargo and rideshare firms transporting passengers in cars in several locations. To date, these vehicles have had a human back-up driver. That's about to change, however, as Waymo (an autonomous car development company) has announced it will begin transporting passengers in Arizona without a human back-up driver.
On March 18, 2018, a pedestrian crossing a street in Tempe, Arizona was struck and killed by a Volvo Uber car that was reportedly operating autonomously with a human back-up driver behind the wheel. Certainly this is a tragedy for the victim and her family.
While it's still early in the accident investigation, according to the Tempe Chief of Police, the preliminary findings seem to indicate that the ridesharing company is likely not at fault for the accident. The police chief stated “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 she came from the shadows right into the roadway.”1 As the investigation continues, Uber has pulled all of its self-driving cars from public roads in Arizona, San Francisco, Toronto and Pittsburgh.2 It remains unclear as to what level of autonomy (e.g., 1- 5) the Uber Volvos have reached. Cars that are deemed Level 2 automation, for example, are not considered autonomous. These are vehicles with some assisted driving - like high-end cruise control. A confidential settlement between Uber and the victim's family has already been reached.
A previous fatality occurred in Florida on May 7, 2016, involving a Tesla operating in driverless mode. It was reported that the Tesla's cameras failed to distinguish a white 18-wheel tractor trailer crossing the highway against a bright white sky resulting in the fatal crash.3 The subsequent accident investigation largely exonerated Tesla's Autopilot system (Tesla's term for its high-end cruise control feature). Further review found that the driver ignored seven audio and visual warnings to place his hands on the wheel - leading to the fatality. The Tesla involved in this accident had Level 2 automation. In its response to the fatal accident described above, Tesla noted, “This is the first known fatality in just over 130 million miles where Autopilot was activated. Among all vehicles in the U.S., there is a fatality every 94 million miles. Worldwide, there is a fatality approximately every 60 million miles.”4
Following the first Tesla fatality, some believed that the federal government would slow down the commercialization of autonomous vehicles. These thoughts are undoubtedly being resurrected. Even before the Uber accident, there was concern in the U.S. Senate, particularly about Level 5 autonomous vehicles (no steering wheel, accelerator or brake pedal). However, some believe autonomous vehicles would result in an 80% - 90% reduction in the frequency of accidents. With the promise of fully autonomous vehicles dramatically improving safety, commercialization may be slowed, but if so, likely only a little.
Note: They will not be perfect.
Even if autonomous vehicles perform as promised some 10% - 20% of accidents could still occur. Keep in mind that humans are responsible for about 90% - 94% of vehicle accidents. In 2016 alone, with humans behind the wheel, 40,200 died in vehicle accidents.5
Development and commercialization of autonomous vehicles is a global phenomenon with companies in Europe, Asia and elsewhere racing U.S. companies to commercialization. Are we willing to concede the technological edge?
It is important to note that currently not all autonomous vehicle technology is created equal. There are reports that the Uber technology was struggling to obtain 13 miles between human interventions,6 while in the past few months Waymo’s driverless tech was able to achieve nearly 30,000 miles driven without human intervention.7
Technology will continue to evolve and improve, but autonomous vehicles will not be accident free. Fatalities will still occur, but maybe, far fewer than caused by human drivers. If fully autonomous vehicles (i.e., Level 5) perform anywhere near expected, auto carrier insurers will see significant reductions in frequency as well as shrinking Personal Auto premium. If fully autonomous vehicles have no steering wheel or pedals, there may be an accompanying shift from the current legal standards for auto accidents (negligence or comparative negligence) to product liability (which, in the U.S., is strict liability). These changes may arrive sooner than anticipated. Waymo has already struck a deal to purchase thousands of Chrysler Pacifica minivans for driverless rideshare services. Subsequent to the Uber accident in Arizona, Waymo announced it will also purchase 20,000 vehicles from Jaguar Land Rover for its rideshare venture. The times they are a changin’.
Endnotes
- "Police Say Uber Is Likely Not at Fault for Its Self-Driving Car Fatality in Arizona," Fortune, March 3, 2018.
- "Uber halts self-driving tests after pedestrian killed in Arizona," The Verge, March 19, 2018.
- "Tesla driver dies in first fatal crash while using autopilot mode," The Guardian, June 30, 2018.
- “A Tragic Loss,” Tesla, June 30, 2016.
- “U.S. Traffic Deaths Rise for a Second Straight Year,” The New York Times, February 15, 2018.
- "Uber’s Self-Driving Cars Were Struggling Before Arizona Crash," The New York Times, March 23, 2018.
- "The race to build a self-driving car charted”, Quartz, March 3, 2018.