Comma.ai’s run-in with the U.S. traffic safety body has revealed weaknesses and unpreparedness of all parties involved in the still-embryonic autonomous car initiatives.
In response to safety concerns by regulators in the United States, commai.ai founder George Hotz has cancelled the launch of his first product—an aftermarket ADAS unit priced at $999.
The announcement has resulted in autonomous vehicles privately freaking out. Many advocates of artificial intelligence and autonomous vehicles see Hotz’s hot-blooded withdrawal as “major news.” They worry that it could sideswipe the future of highly automated vehicles.
The fuss started last week when the U.S. National Highway Traffic Safety Administration (NHTSA) sent Hotz a letter seeking answers to questions regarding the functionality of his device. The agency also noted that a failure to respond to their outreach would lead to fines of $21,000 a day.
Reverting to social media, Hotz tweeted last Friday: "The comma one is cancelled. Comma.ai will be exploring other products and markets. Hello from Shenzhen, China."
"Would much rather spend my life building amazing tech than dealing with regulators and lawyers. It isn’t worth it," he said in another tweet.
Comma.ai’s run-in with NHTSA has revealed weaknesses and—more to the point—the unpreparedness of all parties now involved in still-embryonic autonomous car initiatives.
This controversy exposed, first of all, the cavalier attitude of an upstart start-up with little respect for prudent regulations. It highlights the unpreparedness of regulators to deal with newcomers—neither OEMs nor tier ones—to the hidebound auto industry. It suggests strongly that NHTSA has few plans for approving aftermarket ADAS products. Nor does it appear to have formulated ways to certify ADAS/autonomous car technologies applicable both to start-ups and established companies.
A widespread response in the industry was to deem Hotz’ tweets adolescent and self-absorbed. At best, in this view, Hotz has offered a knee-jerk reaction to legitimate authority. Most alarming, however, to many adults in the autonomous vehicle community wasn’t Hotz’s impetuosity. They fear, rather, the deeper implications of his action.
His industry critics say that Hotz saw the game not going his way. He got upset and left the field, taking the ball with him and ending the game for everyone.
Not good either for the industry or for the market.
Ravi Puvvala, CEO of Savari, a leading supplier of DSRC-based V2X technology, told EE Times, “This is going to be most debated topic in the industry.”
Sateesh Kumar, a former innovation and incubation executive at Cisco, serving on the advisory board to the Alliance for Transportation Innovation (ATI21), told EE Times, “I don’t think this [comma.ai] incident should discourage anyone” from attempting to offer innovative products.
By the same token, nobody—including innovators like comma.ai’s Hotz—should feel that they’re free from reasonable oversight, Kumar added.
ATI21, a consortium of transportation technology innovators, subject-matter experts and researchers, strives to serve as the mediator between eager start-ups and government.
For any autonomous driving technology to take hold, four key aspects must be present and working hand in hand, Kumar said. “First, evolution of the technology, second infrastructure–such as V2X–which is very important, third, public policy for safety and reliability, and fourth, trust and confidence of consumers.”
The failure to coordinate these obligations will put the future of autonomous driving at risk, Kumar emphasised.
EE Times asked Kumar if regulators are serious about encouraging more innovation, what NHTSA could have done differently in the comma.ai case.
Kumar finds the key issue in NHTSA’s rather intimidating threat to fine the start-up $21,000 per day if comma.ai fails to prove its product’s safety of the product.
Kumar said, “I believe, NHTSA should have taken little softer approach. Rather than scaring [Hotz] with a fine, they should have asked if comma.ai can work with NHTSA and car companies to ensure compliance with safety guidelines along with tighter integration with ADAS.”
Savari’s CEO Puvvala offered a similar sentiment: “In my humble opinion, NHTSA should have set up a certification process before they started levying fines on start-ups. This is unfair to levy fines without a formal process to certify AV based aftermarket products.”
But here’s the thing.
There is no such thing as a certification process in the automotive industry—in the past or present. Auto making is one of the few U.S. industries in which federal regulators are powerless to require manufacturers to go through a third-party certification process before launching new products. Toasters and hair dryers are more tightly monitored than cars.
This anomaly turns even more problematic in the aftermarket. The agency, for example has yet to issue any rules of the road for add-on ADAS products.
In its letter to Hotz, NHTSA stressed that comma.ai must “take appropriate measures to ensure the safety of your product and its foreseeable uses.” But nowhere in the document did the NHTSA spell out its idea of an “appropriate measure.”
Asked about that, Roger Lanctot, associate director in the global automotive practice at Strategy Analytics, told us, “NHTSA isn’t really sure what ‘appropriate measures’ are.”
He added, “They don’t yet exist. Hotz essentially was put in the position of being the poster child or test pilot for all future aftermarket companies.”
In Lanctot’s opinion, “Hotz wanted no part of that–so some future Hotz-like entrepreneur will have to emerge to test the NHTSA’s willingness to allow aftermarket solutions. In the grand scheme of things–with 250 million cars already on the road equipped with little or no self-driving tech–the prospect of an aftermarket solution is overwhelmingly tempting.”
It’s important to note that NHTSA offers no certification process—both for finished cars and aftermarket products—today.
But here’s the problem. Lanctot pointed out, “Like security, self-drivability is almost impossible to ‘certify.’ This is why Volvo has said they will take responsibility for liability when they finally ship cars with full automation. Just as was described at the California DMV hearings, the only path to market is ‘self-certification.’”
Lanctot sees this “very treacherous territory for regulators.” On one hand, “they absolutely want autonomy,” he said. But on the other, “they can’t certify it.”
Judging from the many questions NHTSA asked comma.ai about the start-up’s aftermarket ADAS unit, the agency is struggling to figure out what the unit does.
In the Special Order letter, the NHTSA asked comma.ai to “describe in detail how the comma one is installed in a vehicle and provide a copy of installation instructions for the comma one.”
The NHTSA also inquired, “Describe the functionality of comma one, if any, if installed in an unsupported vehicle.”
NHTSA’s question went further into product details, asking comma.ai to “describe in detail how the comma one impacts a vehicle’s rearview mirror, including whether it requires removal of the rearview mirror or the extent to which it blocks or obstructs the rearview mirror.”
None of these are unreasonable questions, but this line of inquiry also makes it appear that NHTSA, without a clear guideline for aftermarket products, is groping in the dark.
In the end, the comma.ai incident exposes the contentious intersection between Silicon Valley companies with untested technologies and incumbent car OEMs and tier ones who may have already won the trust of NHTSA. The hitch, however, is that these old-line companies now also have untested autonomous-driving technologies.
NHTSA told comma.ai, a start-up, that “it is insufficient to assert, as you do, your product ‘does not remove any of the driver’s responsibilities from the task of driving.’”
On the other hand NHTSA routinely accepts, without challenge, the safety claims of established car OEMs and tier ones when they launch new products. No delays or fines are imposed or threatened.
Asked if this is a double standard, Lanctot said, “Of course it is a double standard.”
In his view, “Hotz created the problem because he was doing everything possible to hit a ‘mass market’ price point which was truly terrifying for the regulators.”
Lanctot also added, “I am sure NHTSA was under immense pressure from car makers, safety advocates and others to set up some kind of hurdles, at least, in the path of comma.ai. Hotz could have been and could still become the point of the spear for aftermarket automation—walking the regulators through his development and testing activities. But, for Hotz, this probably looked like an expensive process leading to a dead end.”