Friday, March 24 was clear and sunny in Tempe, Arizona, with a high temperature kissing 78 degrees – perfect weather to end the work week. For the ride-hailing zealots at Uber, a beautiful sunset was a harbinger of much ugliness to come. In the southwestern dusk, an accident involving one of Uber’s autonomous Volvo XC90 test fleet cars happened. It’s been the crash that keeps on giving.
The news goes incandescent when something bad happens to Uber or autonomous vehicles. A single story with both an autonomous car crashing and Uber is like double Christmas for the profit-centered, sensationalized news media. This time, there were even pictures of the overturned Uber vehicle.
Everyone please chill
I wish we were at the point where every autonomous car incident didn’t result in a flurry of headlines and wide, shallow coverage. I’m no fan of Uber, the company is clearly rotten from the head down. Nor am I enthusiastic about ceding driving responsibility to the robots, but must every collision be reported like the return of the Zodiac Killer?
What did you all expect?
Human drivers can’t get through a day without crashes, so it’s unfair to assume the emerging autonomous technology will be perfect right away. There are bound to be more incidents as this stuff is refined. A picture and a headline don’t tell the whole story, and this situation can at least give a rich set of data for improving algorithms, sensors, and software. It’s the machine learning equivalent of a teachable moment.
The photography from the scene is dramatic, but there were no injuries. The police report describes a traffic situation that may have made the autonomous technology blind to a car turning left across an oncoming lane. An attentive human driver would likely have backed off. Instead, eyewitnesses recalled the Uber vehicle either sped up to get through the yellow light, or at least failed to recognize the complex and uncertain situation it was barreling toward.
What to do?
A “smart” car ignorantly driving itself into trouble is a problem. Software can easily be made more cautious in uncertain conditions. A car programmed to speed up to make it through the yellow light before it changes to red could probably be really, really good at using speed/distance/time calculations to beat the light. Of course, that would be a hugely irresponsible choice. On second thought, this is Uber we’re talking about, so let’s take a look at that. Machines are only as deadly as we tell them to be, after all.
Cars are already getting more cameras and sensors, and more advanced technology such as LIDAR sensors keep dropping in price. Situational awareness is a critical piece of what flesh-and-blood drivers provide, and technology hasn’t yet caught up. Vehicle to Vehicle (V2V) and Vehicle to Infrastructure (V2I) communications will play a big role in preventing this kind of crash and should probably be mandatory at some point to give the self-driving car knowledge of its surroundings. The programming part will happen well before the Vehicle-to-anything communication part gets off the ground. You can buy a Cadillac today with vestigial V2V, but it’s going to take at least a couple years to achieve any sort of critical mass.
It’s not the machines, it’s the people.
Despite the work that’s still left to do, it’s not the fault of the technology. Even the relatively primitive semi-autonomous driver assistance technology you can get off the showroom floor today is brilliant at preventing crashes. This stuff does its job.
It’s so good, in fact, that it’s already easy to abuse. The temptation to pick up the smartphone absolutely skyrockets while the car is handling the stop-and-go on its own. Tesla Autopilot has sparked its own sub-category of people-should-know-better web videos. I’m making the “guns don’t kill people…” argument, and it’s just as true in this instance. People acting irresponsibly is the major downfall of this advanced technology, and there’s no algorithm for that.
The technology is only going to get better, but…
The problems will be solved, but not without more mistakes by confused computers. In the case of Uber, the problem isn’t technology, it’s Uber. Plagued by scandals caused by an internal culture that’s hostile to women, playing fast and loose with personal data, and intentionally flouting laws, Uber has a ridiculous valuation of more than $60 Billion. This is a company led by Travis Kalanick, a self-admitted man-child. A company which doesn’t even make the product that makes its existence possible (cars) is absurdly considered more valuable than real manufacturers.
Investors are drawn to the “disruption” Uber has caused, its fast growth, and wide adoption by users. In some circles, “to Uber” has attained verb status, a mark of success in its own right. Remember how quaint we all sounded when we’d say “do a web search” instead of “Google it”? Industry watchers, on the other hand, don’t put that much stock in Uber. The Navigant Research Leaderboard Report on automated driving puts the company way behind most others working on autonomous technology and platforms. The report is a snapshot in time, but it’s informed by the weight of history and an informed look at who’s doing what.
Uber has partnered promiscuously, working with Volvo, Daimler, and investors including Toyota and GV (Google Ventures), the venture capital side of Alphabet. The company is developing its autonomous technology to eventually get rid of those pesky human drivers who need background checks, salaries, and protection from exploitation. Despite the strong bonding moves, market leadership, and ongoing technology development, Navigant ranks Uber behind other companies doing this mobility work. That’s because the research firm feels the future favors vertical integration, and automakers are much closer to that ideal already. Then there’s the behavior…
Grow up or get out
“Uber has a history of seeking to sidestep regulations covering taxi services,” explains Navigant. “While this strategy has allowed it to expand its business rapidly, this desire to operate outside regulatory oversight could also put it in the crosshairs of regulators concerned about automated vehicles,” the report notes. In other words: the party may be over soon.
Employees keep fleeing Uber as the weight of endless controversy gets heavier. That stuff will eventually take its toll, dinging the company’s valuation, forcing Kalanick out, or some other as-yet-unimagined drama. Even Arianna Huffington, she of deep skill extracting value from others’ hard work, won’t be effective in her role of straightening Uber out unless she plays some very aggressive hardball. Or maybe the behind-the-scenes money shenanigans will annoy investors, too. There’s lots of grist for the theory mill.
The other point Uber ought to pay attention to is the strong uppercut automakers are poised to deliver. Automakers build cars from scratch. Future-looking car companies will buy or partner with ride hailing companies. In fact, they already have. While Uber has a high make-believe value and leading product, automakers have profitable operations. They can wait, or buy, Uber out.
Maybe that’s the end game for Uber: sell itself while it’s still a dominant force. In that case, it really doesn’t matter what automakers do, Uber isn’t in it for the long game. Wait too long or ask too much while the self-inflicted wounds keep piling up, and Uber may be signing its own death warrant. After all, it’s a scheduling and light-duty logistics app – not exactly hard to re-create.
In the meantime, Uber doesn’t seem intent on cleaning up its act. Peeling back the layers just reveals more poor behavior, which has made it difficult for consumers who care about ethics to choose the service no matter how good it is. Just look at the #DeleteUber movement if you need evidence.
My biggest take-away from this latest report by Navigant is this: If Uber doesn’t succeed in driving itself into the ground, other companies will.
Dan Roth leads two lives: advertising Creative Director and automotive journalist. He hosted the Autoblog Podcast for 10 years and now co-hosts Wheelbearings.media with Sam Abuelsamid.