Skip to content
no wheel, mo’ problems

Congress debates allowing tens of thousands of cars with no steering wheel

GM needs help from Congress to spin up production of smooth-dashed driverless cars.

Timothy B. Lee | 144
GM hopes to start manufacturing this car, with no steering wheel, gas, or brake pedals, next year. Credit: General Motors
GM hopes to start manufacturing this car, with no steering wheel, gas, or brake pedals, next year. Credit: General Motors
Story text

In January, General Motors unveiled the Cruise AV, a car designed to have no steering wheel and no gas or brake pedals. It seems like science fiction, but GM is completely serious about the project. On Wednesday, GM announced that it is investing $100 million in manufacturing facilities for the new car, with a goal of introducing a commercial taxi service using the vehicles by the end of next year.

But how quickly these cars actually show up on our roads will depend on Congress. Current safety rules require every car to have a steering wheel and pedals—making a car like the Cruise AV illegal without a special exemption. The rules can be changed by regulators, but the rulemaking process typically takes years to complete. So car and technology companies have been lobbying Congress for an expedited process to allow tens of thousands of vehicles like the Cruise AV on the roads ahead of a full rewrite of the safety regulations.

The House of Representatives easily passed a version of this legislation, called the SELF DRIVE Act, last September. But the Senate’s companion legislation, known as the AV START Act, has been making slow progress. With midterm elections looming, insiders say the next few months are crucial. If the Senate doesn’t pass the AV START Act soon, large-scale manufacturing of vehicles like the Cruise AV could be pushed well into the next decade.

Critics of the legislation say that wouldn’t be a bad thing. They argue that rulemaking is slow and deliberate because that’s what’s required to protect public safety. They point out that there are plenty of avenues for companies to test driverless car technology under existing rules in the meantime.

But advocates warn that delaying the introduction of driverless cars could easily cost many more lives than it saves. Errors by human drivers lead to crashes that kill tens of thousands of Americans every year. Driverless cars may be able to prevent many of those deaths. So slowing progress on driverless car technology—even by a year or two—could cost thousands of lives.

“The rate at which the public will realize the benefits of fully autonomous vehicles will be delayed if we have these regulatory barriers in place,” says Greg Rogers, an analyst at the energy policy group Securing America’s Future Energy (SAFE).

Waymo abandoned plans to nix steering wheel

The big obstacle to introducing a fully driverless vehicle like the Cruise AV is a set of rules known as the Federal Motor Vehicle Safety Standards (FMVSS). These regulations specify, in excruciating detail, the safety standards that must be met by every major component of an American car.

Many of these regulations were written decades ago, and their authors likely never considered the possibility that a car might not have a human driver at all. As a result, the regulations have requirements that don’t make sense for a fully autonomous car, like steering wheels, gas pedals, brake pedals, controls for turn signals, rearview and side view mirrors, and so forth. And it’s not just that these features are unnecessary—a human passenger unexpectedly grabbing the wheel of a fully autonomous car could easily cause a crash.

Google’s Firefly car had no steering wheel or pedals.
Google’s Firefly car had no steering wheel or pedals. Credit: Google

The first company to grapple with this issue was Google. Back in 2015, the company had designed a small two-seat driverless vehicle, dubbed the Firefly, with no steering wheel, pedals, or other driver controls. The company’s lawyers warned that introducing the Firefly as a commercial product could get Google in legal trouble.

So Google wrote to the National Highway Traffic Safety Administration (NHTSA), the agency responsible for updating and enforcing the FMVSS. Google asked the NHTSA for permission to treat the self-driving software as “the driver” for purposes of FMVSS requirements. The company also asked for permission to ignore provisions of the FMVSS that related to a human driver, arguing that these regulations were inapplicable since a Google self-driving car wouldn’t have a human driver.

NHTSA was OK with treating Google’s software as “the driver” when interpreting the FMVSS. But NHTSA said no to ignoring explicit requirements for human-accessible controls. For example, one rule requires that a brake “shall be activated by means of a foot control.” Google could ask the NHTSA to open a formal rulemaking procedure to change these rules. Google could also request a special exemption from the rules—with some important limitations we’ll discuss below. But the NHTSA said that simply pretending the rules weren’t there wasn’t an option.

This turned out to be a moot issue because the NHTSA’s letter arrived as Google’s driverless car program was undergoing a major leadership change. Google ousted Chris Urmson as head of the driverless car project in favor of auto industry veteran John Krafcik. Under Krafcik’s leadership, Google abandoned the idea of designing its own car from scratch. Instead, the company began negotiating with incumbent car companies—first Ford, then Fiat Chysler—to buy standard human-driver cars that could be modified for driverless applications.

This new strategy has allowed Google—now Waymo—to sidestep the FMVSS issue. Waymo buys standard Chrysler Pacifica minivans that are already compliant with the FMVSS, then outfits them with sensors and other driverless car hardware. Waymo’s plan is to leave the driver controls in the cars but not let anyone touch them.

This will likely allow Waymo to bring a driverless taxi service to market without needing any changes to the FMVSS or special approval from the NHTSA. But it’s obviously a clumsy solution. It means one of the seats on each vehicle is essentially unusable. And Waymo has to worry about passengers—perhaps a rowdy teenager or a passenger who has had too much to drink—grabbing the wheel at the wrong moment and possibly causing an accident.

GM wants to drop the steering wheel altogether

Now GM is following in Google’s footsteps, seeking to build a fully driverless car with no driver controls. And the company has a few different options to do this.

One option, as we’ve already discussed, is to formally petition the NHTSA to change the rules. Federal agencies have to go through a complex process known as “notice-and-comment rulemaking” in order to change existing regulations. GM could make a list of regulations that don’t make sense in a driverless car and ask the NHTSA to initiate a rulemaking process to change those rules.

Sensors on top of a Cruise vehicle.
Sensors on top of a Cruise vehicle. Credit: General Motors

Almost everyone agrees that this will happen eventually. The big problem with it, however, is that it’s slow. One primer on the FMVSS estimates that it takes about five years, on average, to make a significant change. Given the number and complexity of the issues the NHTSA would have to deal with here, it could easily take longer to fully update the FMVSS for a self-driving age. And of course the technology is changing so quickly that there is a risk regulations written this year will be obsolete by the time they take effect around 2023.

GM’s other major option is to petition the NHTSA for permission to build a car that doesn’t fully comply with the FMVSS. Current law already provides an avenue for doing that, but it comes with a couple of significant restrictions.

First, it requires the applicant to demonstrate that the new vehicle will be at least as safe as a vehicle that is fully compliant with the FMVSS. Given the significant limitations of GM’s current test cars, it is not obvious that GM will be able to convince the NHTSA that a car with no steering wheel will necessarily be as safe as a standard vehicle.

Even if GM can win that argument, however, the larger problem is that these exemptions can only be used for up to 2,500 vehicles per year. That’s a lot of cars for testing purposes—Waymo, the industry leader, only has a few hundred cars on the road today. But it’s a small number of vehicles for a company like GM. The Orion plant, where GM plans to assemble the Cruise AV, has built 5.1 million vehicles over 32 years, which works out to more than 3,000 a week. The company isn’t going to recoup its $100 million investment in Cruise AV manufacturing facilities making 2,500 cars a year.

The SELF DRIVE Act, which the House passed last December, dramatically raises these limits. It would give the NHTSA the power to allow a carmaker like GM to build 25,000 vehicles in the first year, 50,000 in year two, and 100,000 in years three and four.

The numbers are a little different in the Senate version, the AV START Act: 15,000 vehicles in the first year, 40,000 in the second year, and 80,000 per year after that. But the basic effect would be the same: car and technology companies would have a lot more room to build serious driverless car businesses without waiting years for the NHTSA to rewrite the FMVSS.

At the same time, AV START requires the Secretary of Transportation to begin rewriting the FMVSS to accommodate autonomous vehicles. She’ll be required to make a list of rules that effectively require a human driver and draft alternative standards that a fully automated car can satisfy.

Finally, the legislation preempts state efforts to regulate the design, construction, or performance of autonomous vehicles. States would still be able to regulate the behavior of autonomous vehicles on public roads, as well as the sales and repair of autonomous vehicles within the state. But regulation of the car itself would be off limits—an effort to make sure car companies don’t get trapped in a maze of contradictory requirements.

Opponents say Congress should take its time

In a polarized and dysfunctional Congress, self-driving car legislation is one of the few things that has a decent chance of passing with bipartisan support. The SELF DRIVE Act passed the House last September with an uncontested voice vote. Supporters were hoping to repeat that feat in the Senate, but the process has not been so easy in the upper chamber.

In a Wednesday letter, five Democratic senators detailed their issues with the legislation.

“We are concerned that the bill indefinitely preempts state and local safety regulations even if federal safety standards are never developed,” the letter said.

The senators argued that “exemptions from current safety standards should be temporary and reviewable.” And they argued that federal regulators should be charged with gathering data about the performance of driverless cars—including information about individual incidents.

The letter is significant because it is signed by Sen. Dianne Feinstein, who has emerged as one of the legislation’s most persistent critics. The Senate’s dysfunctional rules require a long, tedious process to pass legislation that has significant opposition. With many other issues competing for attention, it’s unlikely that AV START will get the necessary floor time. So the best chance for passing the bill is with an expedited process known as unanimous consent.

But that’s only an option if support for the bill is unanimous, and right now it isn’t. It faces opposition from Feinstein and at least four other Democrats. Feinstein has been blocking the legislation despite the fact that some of the nation’s leading autonomous vehicle companies—including Waymo, Uber, and Lyft, as well as GM’s Cruise unit—are in her state.

Sen. Dianne Feinstein (D-CA), left, has emerged as a leading skeptic of the AV START Act.
Sen. Dianne Feinstein (D-CA), left, has emerged as a leading skeptic of the AV START Act. Credit: David Lee

Feinstein and her allies insist they’re not opposed to driverless car legislation in principle. “We would like to work with you to resolve several outstanding concerns,” they wrote. But it’s not clear how long it will take to reach consensus.

The slow pace of progress suits some public safety groups just fine. A letter from 27 public safety groups earlier this month argued that passing legislation in haste “will have disastrous consequences for public safety and public acceptance of driverless cars.” They opposed key provisions of the legislation, including the proposal to raise the exemption cap.

And it’s not clear that GM, the company that has shown the greatest interest in building cars that would need an FMVSS exemption, will actually be ready to mass-produce the Cruise AV by the end of next year. A recent report from the Information found that Cruise vehicles were still struggling to cope with basic challenges like tunnels, construction sites, and even irregularly shaped bushes. Waymo has been working on these problems several years longer than GM has, and it still hasn’t brought a product to market.

On the other hand, as noted above, delaying the introduction of driverless cars could be harmful to public safety as well. Car crashes cost more than 30,000 lives per year in the United States alone, with drunk, drowsy, and distracted drivers accounting for many of those fatalities.

We’re still a fair way away from having driverless cars that are provably safer than human drivers. But there’s every reason to think driverless cars will be safer than human beings in the long run. And reaching that point quickly will cost billions of dollars of investment—investments companies are more likely to make if they see a clear path to bringing the cars to market at scale.

Listing image: General Motors

Photo of Timothy B. Lee
Timothy B. Lee Senior tech policy reporter
Timothy is a senior reporter covering tech policy and the future of transportation. He lives in Washington DC.
144 Comments