Howes: Hype takes back seat in self-driving car death
As lone traffic fatalities go, the autonomous-vehicle accident last Sunday that claimed the life of an Arizona woman is no ordinary collision.
It should force a reappraisal across the industry and everyday consumers, among government regulators and overly bullish investors. The result: a clearer understanding that a self-driving future must reckon with a real-world reality of rules, safety and imperfect technology with a lot to learn.
Uber Technologies Inc., operator of the self-driving Volvo XC 90 that struck and killed 49-year-old Elaine Herzberg, is drawing scrutiny for the effectiveness of its self-driving system. It should, because the autonomous world is not just around the corner.
It’s got a long way to go — and people in the middle of trying to realize the vision know it. Toyota Motor Corp. is idling autonomous testing in Michigan and California, at least temporarily. Competitors are staying mum and observing, because the accident Arizona investigators called “unavoidable” could just as easily have included one of them.
“Events like this can cause setbacks,” said Gov. Rick Snyder, a strong advocate of self-driving vehicle legislation backed by state lawmakers. “When somebody gets killed, it’s clearly important that there be a lot of review and analysis to see what happened and what the causes were.”
And not just by politicians and regulators. Wall Street is questioning the hype that’s driving investor enthusiasm, with at least one analyst suggesting the lawyerly caution of traditional automakers steeped in the interconnected world of regulations and liability might be wiser than the free-wheeling business cultures of Silicon Valley.
That’s especially true when failure isn’t a black computer screen, but a one-way trip to the morgue — and a potential flood of products liability lawsuits looking to assess blame and reap fat payouts. On that score, Detroit might have a thing or two to teach Big Tech.
“Conservatism might be the right approach,” Barclays wrote this week in a note, adding that “the news likely reinforces the challenges ahead” for Tesla Inc. and its approach of using consumers as “active beta-testers.” Probably not a good idea.
The inevitable finally happened. After a Tesla customer driving his Model S in “Autopilot” mode died in an accident investigated by federal regulators, an Uber autonomous test vehicle struck and killed a pedestrian. That’s a chilling reminder of just how long it could take to realize the accident-free transportation system promised by autonomy:
Longer than many of its enthusiasts admit.
“We’re still very much in the early days of making self-driving cars a reality,” Bryan Salesky, CEO of Ford Motor Co. partner Argo AI, wrote in a blog post last fall. “Those who think fully self-driving vehicles will be ubiquitous on city streets months from now or even in a few years are not well connected to the state of the art or the safe deployment of the technology.”
Sensors “have a long way to go,” he added. Cameras struggle in poorly lit conditions, but lidar sensors don’t. And “combining all of this into one comprehensive and robust picture of the world for the computer to process is incredibly difficult. Developing a system that can be manufactured and deployed at scale with cost-effective, maintainable hardware is even more challenging.”
Exactly right. Plus, complex systems designed to drive cars don’t act like human beings. They don’t routinely roll through stop signs, don’t abruptly change lanes without signaling, don’t make an aggressive left turn even as someone with the right of way prepares to turn right. They’re programmed to be perfect.
Each scenario, many of them managed by human experience and intuition, must be learned by the artificial intelligence systems embedded in autonomous vehicles — data collected over millions of miles of testing by Ford and General Motors Co., Tesla and Waymo, the self-driving car unit of Google parent Alphabet Inc.
The technology is advancing, experts say. But ready to move real people to real places? Not yet. Count on a single traffic death in Arizona to have meaningful implications for multibillion-dollar industries pressing to be first to market with a car that drives itself.
Not ready. But not to be ignored, either. And that’s why automakers and their suppliers will keep sending contradictory signals about the future. It’s not about either cars and trucks or self-driving vehicles. It’s about all of the above.
Daniel Howes’ column runs Tuesdays, Thursdays and Fridays. Follow him on Twitter @DanielHowes_TDN, listen to his Saturday podcasts, or catch him 3 and 10 p.m. Thursdays on Michigan Radio’s “Stateside,” 91.7 FM.