Here’s how California plans to regulate driverless cars

Enlarge / An Uber driverless Ford Fusion in Pittsburgh in 2016. Cars like this will soon be legal for commercial use in California.
Photo by Jeff Swensen/Getty Images

Car and technology companies are now just a few years—possibly even a few months—away from launching commercial services built around driverless cars. And state regulators are facing pressure to get ready by clarifying the rules of the road for fully autonomous vehicles.

On Wednesday, California published new draft rules that give us our clearest look yet at how the driverless car industry will be regulated in California. The new rules, which are subject to a 15-day public comment period before they become final, address a wide range of issues. But perhaps the most significant change is that the new rules formally recognize that companies will be deploying fully driverless vehicles, not just testing them.

That means it will likely be legal to operate commercial taxis with no driver in California as soon as next year.

California regulators have thought through many of the practical details that will be involved in turning driverless cars from a science project into a shipping product. The rules require companies to think through details like how driverless cars will interact with law enforcement and emergency workers, as well as how the cars comply with changes to traffic regulations. The rules also want companies to know how they’ll monitor driverless cars, collect crash data, and report crashes to the authorities.

As the most populous state in the union, California has more leverage than most states. Its great weather makes it an attractive place to start using vehicles that aren’t ready to handle rain and snow. Silicon Valley is home to a number of self-driving car startups, and several car companies have also set up research labs in the Bay Area.

So while other states have felt pressure to keep their rules at a minimum to attract self-driving car companies, the Golden State has felt comfortable imposing stronger—or more onerous, depending on your perspective—regulations than most. Yet, for the most part, California regulators are deferring to manufacturers to decide how best to make their vehicles safe and compliant with the law. Carmakers must certify that they have plans in place to address a variety of situations, but the regulations leave the details up to individual car companies.

California’s vision for autonomous vehicles

If the car has no driver, then the driver can’t be held responsible for accidents, so liability is likely to fall to the manufacturer of the self-driving car. So the new rules require companies deploying self-driving cars to demonstrate they have the ability—either through insurance or their own financial resources—to pay out judgments of up to $5 million in case of a crash involving their vehicles.

Driverless cars need to comply with state and local driving laws, and the new regulations require automakers to certify that they will be able to promptly update car software as those laws change. If the car is sold to a customer, the car company has to make the updates available. But the customer is responsible for making sure the updates are applied.

For vehicles that are sold to customers, manufacturers must explain how they plan to train customers on the safe operation of the vehicle.

Most experts expect early self-driving software to only operate in certain environments, e.g., on pre-mapped roads and only when there’s no snow on the road. Fully self-driving cars—defined as “level 3” or higher in industry jargon—will need to either hand control back to the human driver or pull over to the side of the road and stop in unapproved environments or situations. California’s laws require manufacturers to describe how their cars will handle these kinds of situations.

California officials also want to make sure manufacturers don’t mislead consumers by exaggerating how sophisticated their self-driving capabilities are. Tesla received some criticism for sending mixed messages about its Autopilot software, leading some customers to believe that Autopilot was a self-driving feature rather than a driver assistance feature, causing a Tesla driver to pay too little attention to the road.

To make sure this doesn’t happen in the future, California’s regulations state that “no manufacturer or its agents shall represent in any advertising of the sale or lease of a vehicle that a vehicle is autonomous” unless it meets California’s definition of an autonomous vehicle.

The rules also seem to clearly sanction the use of fully self-driving vehicles—potentially ones without a steering wheel or gas pedal. For vehicles that don’t require a driver, the manufacturer has to certify that the vehicle will have “a communication link between the vehicle and the remote operator, if any, to provide information on the vehicle’s location and status and allow two-way communication between the remote operator and any passengers.”

The “if any” in that declaration implies that it will be legal to release a car in California that’s totally autonomous, without a driver in the driver’s seat or anyone monitoring the vehicle’s progress remotely. The rules specify that a car without a steering wheel and brake pedals must comply with federal laws, which, of course, manufacturers have to comply with regardless.

Advocates of minimal regulation are winning across the country

The specificity of these rules drives free-market advocates like the R Street Institute’s Ian Adams nuts. He prefers the approach of Arizona, which has attracted a lot of driverless car research to the state by taking a minimalist approach to regulation. He describes California’s regulations of self-driving cars as “one of the worst.”

Adams points out that some of California’s rules focus on vehicle design and safety issues, a domain that’s normally handled by the National Highway Traffic Safety Administration. If NHTSA issues its own rules in this area in the next few years, those rules will preempt the requirements of California law, which could create a confusing situation for auto manufacturers.

At the same time, it’s important not to overstate how onerous these rules are. For the most part, California’s rules merely require manufacturers to self-certify their compliance with a variety of common-sense requirements: that they’ll be able to interact with law enforcement, comply with traffic laws, record data from crashes, and so forth. In most cases, the details of how carmakers accomplish these goals is left up to the discretion of the carmakers themselves. And the self-certification approach leaves limited scope for California authorities to second-guess the design decisions made by regulators.

So while there’s an argument that some of these regulations are overly specific and proscriptive—and could turn out to be superfluous or counterproductive as the industry evolves—the more remarkable thing about them may be how minimal the substantive regulations are. Carmakers, for example, aren’t required to submit data demonstrating that their cars are safe before they’re allowed on the road. California rules don’t mandate the use of specific safety technologies or techniques. In short, even California’s relatively activist regulators are mostly deferring to carmakers to decide how to make sure their self-driving cars are safe.

Powered by WPeMatico