Uber’s self-driving casualty not the last

History will probably remember Elaine Herzberg for a tragic reason: She’s thought to be the first pedestrian ever killed by an autonomous vehicle.

Herzberg was struck Sunday by a self-driving Uber in Arizona, apparently stepping off a median and into traffic too quickly for the car’s sensors to detect her. Even Uber’s safety driver — the human backup sitting behind the wheel — failed to see her before it was too late.

According to the local police chief, it’s unlikely Uber would be at fault in this incident. This may be correct in a narrow sense. But it also suggests a central challenge for the industry: Humans behave unpredictably.

Self-driving cars crash at double the rate of conventional ones, but the technology itself isn’t to blame. One study of crash data found that “selfdriving vehicles were not at fault in any crashes they were involved in.” That’s because while autonomous cars strictly follow the rules of the road, humans generally don’t. They speed, tailgate, stray across lanes and navigate erratically. People are often befuddled by the slow-moving robots in their midst and sometimes crash into them as a result.

Pedestrians complicate matters further. They sometimes ignore rightsof-way or traffic signals. They might be drunk, daydreaming or texting. Herzberg reportedly didn’t use a crosswalk — something most humans do every day but that robots might not anticipate and might never be able to.

Resolving this dilemma is largely up to the companies that wish to deploy these vehicles. Self-driving cars will get better at anticipating everyday human behavior and responding appropriately. And a world with self-driving cars will be safer than one without them. But driving will never be perfectly safe. The industry needs to manage public expectations, even as it reckons honestly and openly with this tragedy.

If it does, the potential benefits are immense. Eventually, the technology could reduce road accidents and deaths, while also conserving fuel, diminishing pollution and alleviating congestion. It’s likely to boost productivity, mobility, growth and well-being. Self-driving trucks alone could be an economic boon.

Prudent regulation could help. Arizona, pertinently, has styled itself as a kind of Wild West for autonomous driving, with more than 600 test vehicles on the road and few rules to guide them. “Arizona welcomes Uber self-driving cars with open arms and wide open roads,” the state’s governor said in 2016.

Yet a welcoming attitude isn’t enough. Governments must help prepare the public for the dangers and limitations of self-driving cars — especially given widespread mistrust of them — and ensure local infrastructure can safely accommodate them. Rules for operators, meanwhile, need not be unduly burdensome to be effective. A bill passed by the U.S. House which requires automakers to meet certain safety standards but gives them flexibility in how to do so is on the right track.

Ultimately, carmakers are responsible when their products are defective. As with all transportation technology, the development of self-driving cars will be marked with controversy and, occasionally, tragedy. It’s up to the industry to ensure that it’s not all in vain.