Editor’s note: This article is an extended version of an op-ed titled “We can’t trust the autonomous-car industry to self-regulate” that ran in the Seattle Times on March 28.
Electric robo-taxis hold great promise for solving the Puget Sound region’s urban transportation problems but not without the right public policies, including objective standards for vehicle safety. Sadly, no such policies were in place in Arizona on March 18 when an autonomous vehicle owned and operated by Uber struck and killed Elaine Herzberg as she walked her bicycle across the street at night.
Electric robo-taxis hold great promise for solving the Puget Sound region’s urban transportation problems but not without the right public policies, including objective standards for vehicle safety.
An alert human driver or competent robot likely would have seen and avoided Ms. Herzberg. We must await completion of a full investigation by the National Transportation Safety Board before reaching final conclusions, but the tragedy creates new motivation to ensure robo-taxis are safe before letting them roam our streets.
The companies developing self-driving technology tout safety as one of the main benefits. Automobiles killed more than 40,000 Americans last year, a horrible toll. Human fault and frailty caused 94 percent of those deaths, so any technology that could reduce this carnage is welcome. Autonomous vehicles could potentially save hundreds of thousands of lives over time, once they work well. But who decides when they are ready? Who gives them their licenses? It shouldn’t be the companies developing them.
The auto industry has fought sensible safety standards for decades, opposing requirements for safety glass, seat belts, airbags, and catalytic converters. Volkswagen committed a massive global fraud by falsifying tests of the emissions from its diesel cars. The auto industry is at it again, working Scott Pruitt at EPA to roll back Obama-era clean car standards. We can’t trust the industry to self-regulate.
Especially since setting proper safety standards for autonomous vehicles is a hard problem to solve. For all the mayhem on our highways, human drivers now drive close to 100 million miles on average before a fatality occurs. The self-driving car industry has logged around 15 million miles of test rides on public roads and has now inflicted its first casualty while testing. Based on these numbers alone, there is a 16 percent chance that the entire self-driving fleet as tested is just as safe as human drivers and simply got unlucky with an early fatality. There’s a better than 4 out of 5 chance, if the technology’s capabilities are uniform across companies, that it is less safe than human drivers. Companies like Waymo have already argued that their technology works better than Uber’s but haven’t given us any way to prove it.
Each company developing self-driving cars would have to log billions of miles of training rides before they could prove statistically that they are substantially safer than humans. Analysts at Rand have shown that too strict a standard for proving safety performance could substantially delay deployment and cost more lives than a less strict standard that allowed the technology to deploy sooner.
The task of weighing potential benefits and costs over time in the face of uncertainty belongs with the public sector, not the companies racing to develop the technology. Despite the promise of breakthrough drugs, for example, we don’t allow pharmaceutical companies to use their own testing procedures and market a new product whenever they say it is ready. Instead, we require them to prove safety and efficacy to the Food and Drug Administration, through a multi-tiered approval process.
In all 50 states, to get a license to drive, a human must pass tests of vision, knowledge of traffic laws, and skill behind the wheel. In Arizona, the robot that killed Ms. Herzberg just needed an OK from the very company that built it. One week after the accident, Arizona’s governor banned Uber from continued testing but Waymo and Cruise robots continue to operate there with no objective standard for vehicle safety.
The autonomous vehicle industry has pushed for federal legislation that would nationalize Arizona’s “trust us” model. The AV START Act (S. 1885) would exempt from federal safety standards hundreds of thousands of robot vehicles to be sold over the next few years and would ban states from stepping in to improve vehicle safety.
For all the talk of wanting to make driving safer, the industry’s bill attracted heated opposition from a broad coalition of safety and consumer groups, appalled by the attempted end run on public oversight. After a similar bill passed the house, Senate Democrats tapped the brakes and have pushed a more balanced approach.
Sensible rules are essential for road safety—to keep the race for market share from encouraging corner cutting, with more tragic consequences to humans. Besides, a reasonable certification process will help build confidence in robots as safe drivers, accelerating adoption once proven out. Domestic companies might even benefit disproportionately if the new standards keep out international competitors that lack a strong commitment to safety.
The omnibus spending bill approved on March 23 includes $100 million for automated vehicle research and development by agencies, local governments, and academic institutions. The industry needs to partner with leading experts at national labs and universities with access to this new funding to shape a performance-based approach to safety that reduces accidents while enabling continued innovation. A graduated certification model that builds on the principles for licensing human drivers would make a lot of sense.
Governor Inslee issued an ill-advised executive order last year that allows autonomous vehicle testing in Washington by companies that “self-certify.” The state gains little by allowing tech companies to test on public roads and puts motorists and pedestrians unduly at risk. Instead, the Governor and legislators should focus on developing policies to enable the rapid scaling of autonomous electric fleets as soon as we know they work. Our region could benefit immensely from broadscale deployment of electric robo-taxis but not without public policy to ensure that they are safe.
DeWitt White
Aircraft, today, can take off, fly from point A to Point B and land safely, autonomously. However, there is always a well qualified pilot in the seat to take command in case of equipment failure. This industry is monitored by the FAA and very stringent rules are in place to verify compliance.
Software companies are driving the autonomous car movement. There is no FAA like institution to scrutinize designs and enforce standards. These companies do little testing, instead they issue a beta version and let the consumer do the testing. How are you going to fare by having an upgrade sent out in the dead of night to your autonomous car, such that it’s not the same car you parked in the garage the night before?
I say this industry needs an FAA like organization to take control and lay down the rules for implementation.
Doug Parrow
I agree completely with the need for governmental regulation of the industry. I think an interesting and challenging question will become how safe is safe enough. Given the enormous number of people that human drivers kill, what error rate will we accept for autonomous vehicles? I suspect that it will be far closer to zero than what has become an acceptable rate among humans.
Spencer
would be interesting to unearth the number of auto – pedestrian collisions that occur in the U.S. every month — to have an intellectually honest conversation about this topic (is autonomous safer than human controlled), it needs to start with the underlying data. we have to guard against “perceived risk” versus the actual statistical risk. That’s not to say that a regulatory body isn’t needed – it very well might be the correct approach.
UrbanMark
How do you make ‘eye contact’ with a self-driving car to be sure it’s safe to cross in front of it ?
Hu
I agree that we need to regulate automated cars. However, other than increased safety, which is a trait that automated vehicles do not have yet, what are the other benefits of AV’s? As far as I know, automated cars will just further incentivize driving, which increases CO2 emissions. Automated vehicles will also eliminate the jobs of millions of truckers, cab drivers, and bus drivers; taking away the way of life for millions of people without a college education. In addition to this, unless congestion pricing is introduced in our big cities, unoccupied Automated Vehicles driving around to pick people up will take up more precious road space, and worsening traffic conditions for everybody. So what is good about this? And why is congress subsidizing something that will solely benefit private tech companies?
asdf2
They could reduce emissions, simply by hastening the adoption of electric vehicles. Choosing a vehicle that is powered by electricity instead of gasoline means trading off a higher purchase price for lower operating costs – a decision that is far more likely to pencil out for a fleet vehicle that’s on the road 20 hours/day than for an individually-owned vehicle that is going to be parked 95% of the time. It will also lead to much fewer vehicles needed to begin with, so less pollution in manufacturing and disposal. Autonomous robo-taxis will also greatly reduce the amount of parking and parking lots needed in city centers, since the cars will be on the road nearly all day.
Finally, the same technology that makes driverless cars possible may also be used to operate driverless buses, so the operating costs of public transit go down, which means the same amount of tax revenue can pay to run the buses much more often. And, if buses can be made to get their own lanes on congested streets, people will still have incentive to ride them over cheap robo-taxis. Express buses can also be combined with robotaxis to solve first/last mile problems in ways that are prohibitively expensive to do at scale, using Uber and Lyft today.
Widespread Robo-taxis may also finally force the issue of congestion pricing in cities, where, without it, would have been politically impossible to impose.