This Could Be The Biggest Hurdle For Driverless Cars

Tesla Introduces Self-Driving Features With Software Upgrade
A member of the media test drives a Tesla Motors Inc. Model S car equipped with Autopilot in Palo Alto, California, U.S., on Wednesday, Oct. 14, 2015. Tesla Motors Inc. will begin rolling out the first version of its highly anticipated "autopilot" features to owners of its all-electric Model S sedan Thursday. Autopilot is a step toward the vision of autonomous or self-driving cars, and includes features like automatic lane changing and the ability of the Model S to parallel park for you. Photographer: David Paul Morris/Bloomberg via Getty Images
Photograph by David Paul Morris — Bloomberg via Getty Images

Last week, something very unusual happened: a legal opinion by the National Highway Traffic Safety Administration set the Internet abuzz. Why? Because it informed Google that the artificial intelligence used to pilot its autonomous vehicle could be viewed as the “driver” for some (not all) regulatory purposes. But the media mostly ignored the grist of the letter, which consisted of a litany of legal smack downs to Google’s proposed interpretation of federal safety rules.

NHTSA’s message was crystal clear: Washington is in the driver’s seat — policymakers, not technologists are going to call the shots on how quickly companies like Google (GOOG) deploy autonomous cars. Just as it has with seat belts, airbags, emissions technology and electric cars, government will exert a decisive force on the pacing and form of the car of the future.

The major obstacles to autonomous vehicle deployment fall into four categories: cost, technology, consumer acceptance and policy. Since 2012, the industry has been racing forward and making remarkable and unexpected progress on the first three areas.

Over the past three years, traditional carmakers and tech companies have dramatically accelerated the development of sensors, artificial intelligence, and software that’s needed to make cars drive. The amount of money and brainpower being channeled toward resolving these issues is staggering.

For example, Google not only has a small army of engineers and computer scientists working on its car, they have established a massive physical testing presence for testing autonomous vehicles on a former U.S. Air Force base in Atwater, California. Last fall, Tesla switched on autonomous features on tens of thousands of its Model S sedans, tracking their cars autonomous miles, providing massive amounts of data on driving patterns and bringing us one step closer to fully autonomous driving. And in the spring, Uber hired away more than 40 of Carnegie Mellon’s top roboticists, putting them to work on autonomous vehicles.

It’s not just Silicon Valley. In January, General Motors (GM) announced a $400 million investment in the ride service, Lyft, with the intent of building a new generation of autonomous taxis. Last month, Ford (F) announced that it is tripling the size of its autonomous vehicle testing fleet – declaring that we should no longer consider it a car company, but instead a mobility company.

Communications giants and chip manufacturers are also in the game. In January, chip giants Qualcomm and Nvidia both showcased huge research investments aimed at providing mobile computing platforms for autonomous vehicles.

Every year, Americans spend more than $2 trillion on car ownership — which is more than twice as much as the global consumer electronics sector — and the automobile is rapidly transforming into the ultimate – and most interconnected – consumer electronic device. So it makes sense that technology powerhouses like Google, Apple, and Qualcomm are entering the race for autonomous vehicles.

Ultimately, it would be surprising if all that inter-industry investment and intellectual horsepower failed to find a way to combine the sensors, processors and algorithms of the 21st century into a consumer-ready autonomous car – and fairly soon.

Until recently, the logical follow on question would have been: sure, but at what cost? However, early predictions that autonomous vehicles would be prohibitively expensive are melting away. Just a few years ago, a lidar sensor – the Cycloptic eye that measures the distances between objects in the physical world and allows many autonomous vehicles to navigate – might have cost $70,000. Today, that price has plunged by a factor of 10 to less than $8,000. The cost of computing power required to run the vehicle keeps falling at the relentless rate of Moore’s Law.

While autonomous vehicles can use either gasoline engines or electric drives, the falling cost of batteries (about 20% per year) is making the electric option look increasingly attractive, even compared to $2-a-gallon gas. GM’s 200-mile-range Bolt goes into production this year and will cost only $30,000 after tax incentives; adding an autonomous system will be drastically cheaper than adding a taxi driver.

As for consumer acceptance, many have wondered whether people will entrust their safety to self-driving cars. It seems increasingly clear that the answer is yes, and they’re even willing to pay for it. A survey by the Boston Consulting Group last year of 1,500 people in 10 countries found that 14% would be willing to pay a premium exceeding $5,000 for a self-driving car. In Paris, Copenhagen, and Vancouver driverless subways haven’t caused much concern.

The big question for consumer acceptance is whether we’ll still want to own cars individually in the future. Toyota (TM) says yes: the company is investing $1 billion in artificial intelligence research for autonomous vehicles, but is betting that individually owned cars will continue to dominate the market. Yet autonomous vehicles should be a perfect fit for rideshare services, as the investments by Uber and Lyft suggest. Rideshare-oriented autonomous vehicles would spend many more hours on the road per day, so their per-mile costs would be very low – with even bigger savings from the fact that there’s no driver to pay.

Our modeling at Valence Strategic estimates if services like Uber and Lyft used autonomous vehicles, their prices should drop by more than two-thirds to around 25 cents per mile. With a slightly different set of assumptions, Larry Burns, one of the leading technologists in the space, estimates a 75% cost reduction to about 15 cents per mile driven. Either way, far from being expensive, autonomous cars operating in a rideshare mode are likely to provide the cheapest per mile transportation ever known to man. It’s a proposition that’s hard to refuse.

And so policy remains as the last major hurdle. As with other fast-moving innovations, policymakers and regulators are struggling to keep pace. The strategic implications for businesses are manifold. Companies intent on capturing part of this emerging market need to understand that, unlike the internet, autonomous vehicles are entering a heavily regulated industry. And NHTSA’s response to Google, demonstrates the challenges of breaking into the automotive space. It’s not just the big questions like: Where and how will these vehicles be allowed to drive? Will they require a special license or manual override? And who will be liable in the case of a crash? It’s also the maddening details like “does the car need to allow for manual control of high beam headlight by occupants?” that will also have to be answered. Silicon Valley bristles at the thought of these niggling regulations holding them back.

Unfortunately, only government can provide the rules of the road and there are many ways that this process could take the entire industry off track. Three stand out. First, policymakers might falter as they struggle with new and unfamiliar questions. (The Federal Communications Commission, which regulates wireless spectrum, is still haggling with the car industry over airwaves it designated for automotive use in the 1990s.)

Well-intentioned, but badly executed rules could stymie progress. For instance, California’s Department of Motor Vehicles recently published rules on autonomous vehicles that infuriated many within the innovation community by requiring a special license to operate autonomous vehicles and also that cars maintain a steering wheel, among other things.

Another threat is that incumbents could intentionally manipulate policymakers and the Rube Goldberg machine of federal and state rules to block the introduction of autonomous vehicles.

Perhaps the most likely and ironic possibility is that a misstep by autonomous vehicle developers could create a public backlash that slams the brakes on autonomous cars. It wasn’t long ago that concerns over fires from lithium ion batteries threatened the deployment of electric cars; a small number of crashes involving autonomous vehicles could similarly have an outsized impact on public perception of their safety and lead to reactionary policies.

One recent article in Bloomberg included thrilling and terrifying footage of an infamous Silicon Valley hacker taking his home-built autonomous car on its first drive. It is easy to see how a deadly crash brought on by a partially autonomous Tesla or one of these enthusiasts could tip the scales of public opinion toward conservatism. Much will depend on chance.

And that is the paradox of policy. Of the four challenges mentioned above, it is the most under our control, but also the most subject to the highly irrational elements of our legal system, politics and society. Bad policy could bring autonomous vehicles to a screeching halt, or be a drag on development. Good policy could propel the industry forward and save millions of lives (over 30,000 Americans die a year in automotive crashes).

The Department of Transportation (DOT) has announced a plan to invest $4 billion in driverless cars over the next decade and the National Highway Transit and Safety Administration (NHTSA) intends to outline a national set of rules and guidelines for deploying autonomous cars within six months. This is a good start. National guidelines will help preempt the proliferation of fragmented state rules that could transform the regulation of autonomous vehicles into a sort of race for the bottom – where states recklessly vie with each other to become unregulated test hubs.

But NHTSA’s response to the Google was also a jarring reassertion of regulatory power. Why jarring? Because the federal government has a relatively poor track record of managing innovation in the auto sector. Policy is often slow, frequently redundant and rarely straightforward. For instance, fuel economy regulations for cars were first enacted in 1975. Despite huge technology shifts within the industry, they were not raised until the passage of the Energy Independence and Security Act of 2007. Even today, three complicated overlapping systems administered by the Environmental Protection Agency, Department of Transportation and the California Air Resource Board govern U.S. fuel economy and greenhouse gas emissions.

Washington will have to do better when it comes to autonomous vehicles. DOT and NHTSA need to create a framework that is flexible but robust, and which takes into account both the possible threats and huge potential benefits of autonomous vehicles. They should also think about the technology’s effect on the environment and greenhouse gas emissions, forcefully nudging autonomous vehicles toward electrification.

Speed is the pride of Silicon Valley. But at the end of the day, Washington will set the speed limit for innovation. Good policy can propel America ahead in the global race for autonomous cars or it can hold it back. So don’t let your driver’s license expire just yet. Because without Washington, autonomous vehicles will go nowhere fast.

Levi Tillemann is the author or The Great Race: the Global Quest for the Car of the Future and managing partner at Valence Strategic. Colin McCormick is a former advisor to the U.S. Department of Energy and Chief Technologist at Valence Strategic.

Subscribe to Well Adjusted, our newsletter full of simple strategies to work smarter and live better, from the Fortune Well team. Sign up today.

Read More

Great ResignationClimate ChangeLeadershipInflationUkraine Invasion