Michigan Engineering News

A street intersection

How we roll

Going driverless could be the biggest reboot of the American transportation system since the horse and buggy went the way of, well, the horse and buggy. Are we ready to make it happen?

Even if you haven’t seen a driverless car on the road, you’ve seen plenty of them in the headlines. Flocks of them are predicted to swarm the streets, ready to roll up at the tap of a smart phone. They’re predicted to be clean, safe, convenient, and if the Google car is any indication, adorable. And they’re predicted to be ready for American streets before you can say, “Fully autonomous self-driving vehicle.”

But a prediction won’t schlep your kid to soccer practice. And getting our streets (and our heads) ready for driverless transportation is going to take a lot more than a cute new car and a can-do attitude. When the cars themselves are ready for prime time, we’ll need to make sure the country that invented car culture is really ready to hand over the steering wheel. We’ll need to make changes to our roads and infrastructure, too, since driverless vehicles will almost certainly require a real-time data connection to get from A to B. In fact, going driverless could be the biggest reboot of the American transportation system since the horse and buggy went the way of, well, the horse and buggy. Are we ready to make it happen?

A city skyline with a green filter

Handing over the wheel

We’re fascinated by driverless vehicles – at least in theory. The most popular videos of Google’s driverless cars have scored tens of millions of views on YouTube. New studies and articles come out literally every day. And according to a recent University of Michigan study, a solid majority of Americans have at least some interest in owning or using a driverless car.

But for driverless cars to become a viable technology, significant numbers of us will need to use them, whether we share or own. And “some interest” may not be enough to make that happen. The same U-M study showed that most Americans aren’t willing to pay extra for a driverless vehicle. And those who are would only pay a premium of about $2,000, on average. Anyone who wants to own a driverless car will need to pony up a lot more than that. Some driverless users will share, of course, but Brandon Schoettle, a co-author of the study and a researcher at the University of Michigan Transportation Research Institute (UMTRI), says that convincing significant numbers of us to make the switch could still take decades.

Carmakers, after all, have spent the last hundred years convincing us that owning and driving cars is about more than just transportation. They’ve positioned human-driven cars as symbols of independence and freedom, an inextricable part of the American experience. How else to explain the fact that a replica of the 1977 Pontiac Trans Am featured in the car-centric Burt Reynolds film “Smokey and the Bandit” recently fetched $450,000 at auction?

Clearly, car culture still has a grip on us. But many marketers believe it’s beginning to wane, particularly with Millennials. Indeed, the group of Americans born between the early 1980s and the early 2000s may be the early adopters that will push driverless toward the mainstream.

“I think driverless technology is going to sell itself to the Millennials,” says Marcie Brogan, founder of metro Detroit-based ad agency Brogan & Partners and co-founder of Ignite Social Media. Both companies have worked with carmakers for years and have extensive experience marketing to Millennials.

“Millennials aren’t getting driver’s licenses or buying traditional cars at the same rate as previous generations,” she said. “They’re less interested in owning, more interested in sharing and more trusting of technology than any generation before them. It’s hard to imagine a group of customers that’s more suited to driverless transportation.”

Brogan believes that simple economic reality may play a role in the adoption of shared and driverless technology as well, as a combination of changing priorities and increasing income disparity puts car ownership out of reach for many in the younger generation. And she believes that a little savvy marketing could spread Millennial enthusiasm for driverless transportation to older drivers, who may be more reluctant to change the way they think about cars and driving.

“The auto industry has always been about selling an experience,” she said. “The experience of owning a car, of driving a car. And I think the people behind driverless cars – whether it’s an auto company or a sharing company like Zipcar – have an opportunity to sell the experience of having a great car that you don’t have to drive, and may not even have to own.”

She says the key to winning over the parents of the Millennial generation may be to show not just what generation may be to show not just what driverless can do for them, but what it can do to keep their children and the environment safe. It’s a relatively new message that she says is already making inroads in the marketing of traditional cars. She cites Subaru’s “Love” campaign, which focuses far more on people than cars.

“The ‘Love’ campaign is a really interesting example of a different approach to selling a car. They’re marketing behavioral change, doing something for your family, for the environment. That’s a very powerful motivator.”

Still, for a certain segment of consumers, safety and responsibility won’t seal the deal. It’s all about the car. And carmakers have a plan for bringing those customers into the driverless fold, too. They’re introducing automation slowly, with technology that enhances the driving experience rather than replacing it. Drivers are already seeing the first of these features, in the form of add-ons like automatic lane keeping and adaptive cruise control. Automatic lane keeping alerts drivers when their car veers out of its lane and can even make steering corrections to stay on course, while adaptive cruise control monitors the car’s speed to maintain a safe following distance behind the car ahead. Features like these are just the beginning, according to Peter Sweatman, the director of the U-M Mobility Transformation Center (MTC).

“We call these day-one values,” he said. “They’re features that drivers can interact with and get value from every day, and they tend to get consumers more excited than crash avoidance. You can’t see the value of safety day in and day out, but if your car can show you the closest parking space or turn a light green at an empty intersection, that’s something that tangibly shows the value of automation.”

Sweatman predicts that the new features will begin to change opinions about automation even before completely autonomous vehicles are rolled out to the public.

“We need to package and present these technologies in such a way that will keep drivers delighted in addition to keeping them safe,” he said. “I think drivers are going to like the sense that they’re no longer on their own, that the system is making the driving experience better. Those experiences are already laying the groundwork for wider acceptance of driverless cars.”

Safety concerns

But it’s not just about marketing. Respondents to the U-M survey expressed a number of practical concerns about driverless technology as well. Safety and liability topped the list, with a respective 82 percent and 77 percent of U.S. respondents expressing concern about those two issues.

The safety fears may be particularly irksome to driverless car developers, since driverless vehicles will almost certainly be far safer than human-driven cars once they make it to market. In anticipation of this, driverless carmakers like Google are already working to change perceptions. And the most powerful tools for doing that, says Schoettle, may be the purpose-built driverless cars that are already appearing on public roads. These cars are built from the ground up to be driverless, in contrast to other test vehicles that retrofit driverless technology to an existing vehicle.

“The car itself is what’s going to build confidence for users,” Schoettle says. “You need a track record that shows that these things will do what they say, where we can say we’ve done X number of accident-free miles on public roads all over the country.”

That may be one reason why big players like Google are working quickly to get purpose-built driverless cars on the road, even if their capabilities are limited. Google’s purpose-built driverless vehicle, for example, is limited in its capabilities: it’s small, and its speed is limited to 25 miles per hour.

But the prototype has one important advantage: it’s cute. It’s also immediately recognizable as a driverless car. And its slow speed makes it seem safe and non-threatening. So even if it’s not the fastest car on the road, it’s a powerful marketing tool.

“Those cars may be one of the most important components to the psychological effects of getting people comfortable with driverless,” Schoettle said. “I think Google’s vehicles are also a great example of how to slowly but surely build a track record of safety.”

THE BLAME GAME

Liability has been another thorny issue for driverless cars. If a driverless car crashes, who’s responsible? Is it the driver? The carmaker? What if a roadside sensor failure caused the crash? In the most litigious society on earth, these are big questions.

Experts like Eric Williams, counsel for the Insurance Institute for Highway Safety, predict that the manufacturers (the ones who design the autonomous systems in the first place) will eventually shoulder more of the liability for crashes.

“The more control the vehicle has, the more decisions you’re shifting from driver to designer. So the designer may assume more liability tomorrow than compared to today,” he said. “There will be lots of uncertainty and growing pains ahead, and that’s why it’s important to get the right parties – the regulators, manufacturers and insurers – talking to each other now.”

Insurance and creativity don’t always play well together. But a little creativity may be just what’s needed as we transition to an insurance system that can accommodate driverless technology. Schoettle suggests that insurers may need to treat auto insurance more like homeowner’s liability insurance: the car owner is responsible for the vehicle even though it’s agreed that the accident may not have been his or her fault.

“If someone trips and falls at my house, it’s not my fault. And yet, I have liability insurance to pay for any injury or damage,” he said. “This is a widely accepted principle in homeowner’s insurance, and I think we could see it applied to auto insurance as well.”

Schoettle also predicts that carmakers could begin bundling auto insurance into the price of cars. It may sound like a radical idea, but he says it could be a clever way to allay several fears about driverless cars in one fell swoop.

“If a car company says ‘We’re so confident in our system that we’ll assume the liability for a crash,’ that’s a pretty powerful statement,” he says. “You’ve just solved the liability problem, you’ve put your money where your mouth is when it comes to safety and you’ve also laid out an attractive marketing proposition. That kind of thing could really change the game when it comes to consumer acceptance.”

A city intersection with a green filter

Building smarter streets

Carmakers and others are spending big money to get driverless cars ready for our streets. But what will it take to get our streets ready for driverless cars? Most experts agree that driverless cars will need to be able to communicate with infrastructure, like traffic signals, in real time. That data stream will require a wireless communication system that’s standardized nationwide – maybe even continent-wide. And we’ll need a system to move all that data smoothly and seamlessly.

IN THE CLOUD

It’s clear that driverless cars will generate an unprecedented amount of data. The connected cars that are being tested today generate about 10 pieces of data per second, per car. And that number will only grow as systems get more sophisticated. Figure in the fact that just maintaining our roads as they are can be a major challenge, and it’s clear that we’ve got a big job ahead of us.

The good news is that all the commotion about driverless technology seems to be generating the will to act – and it’s also creating new partnerships between government, academia, the private sector and others. The U.S. Department of Transportation (DOT), for example, has put up $50 million for the Connected Vehicle Implementation Pilot program, which funds state-based programs to help chart a course to a national system. The goal is to build a system that can work across different vehicles, modes of transportation and government entities. DOT awarded the program’s first grants in September, and deployments are projected to be complete by 2020.

Michael Pack, a researcher at the University of Maryland’s CATT Lab, has worked with several states to build proposals for the DOT program. Established in 2002, the CATT Lab works with connected vehicle and infrastructure data, finding ways to capture, transmit and analyze the huge amounts of data that a connected infrastructure will create. The lab already collects 5.5 billion pieces of transportation-related information every day from state and local agencies – everything from traffic signal data to accident information, parking data and weather.


(BETTER. FASTER. CHEAPER?)

It’s much too early for any firm estimates of how much a connected infrastructure for driverless cars might cost. But in the long run, some experts believe that it could actually save money by replacing the aging, maintenance-intensive patchwork of traffic control systems that we have now.

That’s largely because today’s traffic control systems rely heavily on 20th-century technology. Many intersections use magnetic loop sensors embedded in the pavement. They sense when a vehicle rolls up to an intersection and adjust signal times accordingly. And some intersections still rely on mechanical clocks ticking away inside control boxes. This technology is expensive to maintain and it doesn’t always do a great job of keeping traffic moving.

“The system on today’s roads was mostly designed in the 1960s,” Liu said. “We spend a lot of money maintaining it. On the other hand, the radio units we’re testing are not especially expensive, they’re very reliable, and we’re finding that we don’t need as many of them as initially projected.”

Liu says that a radio-based system that receives signals directly from vehicles could get the job done better, with fewer failures and at a lower cost. The system he’s testing at Mcity has fewer moving parts, which means lower maintenance costs. And because signal timing can be adjusted remotely, there’s less need for crews to go out on the road.

Radio-based systems could also manage traffic flow much more efficiently than older methods. And making vehicles an active part of the system actually increases reliability by adding redundancy to the system – even if a few vehicles don’t have sensors or aren’t working properly, the system will still receive a steady stream of data from other vehicles that are working as intended.

Further in the future, Liu envisions a day when we won’t need traffic signals at all, since human drivers won’t be part of the decision-making process. It’ll be many decades before that happens, but when it does, infrastructure costs could be reduced even further.


Pack says one of the biggest challenges of implementing a connected infrastructure will be simply finding a way to quickly move all that data through the system. At 10 or more records per second, per connected vehicle, the data adds up fast, he says.

“The data stream from connected vehicles quickly gets larger than any data stream we’ve seen for traffic systems,” he said. “The issue isn’t can we grab it and store it, the issue is communications backbone out on the street.”

Government officials and automakers are making progress, having agreed on the basic framework for a set of nationwide connected infrastructure standards. They’ve set a standard language (called the SAE standard) for those 10 records per car, per second, as well as a standard transmission method (called the iEEE standard).

But Pack says that these basic standards won’t be enough to create a connected infrastructure that gets the job done. He says that connected vehicles can generate a wealth of information – everything from outside temperature to air bag deployment, headlight and wiper use, and much more. It’s all valuable, but there are no standardized formats or collection methods for the vast majority of it. If governments don’t get more involved soon, they could end up dealing with a patchwork of data systems, or worse, having to purchase data from automakers.

“I think the DOTs need to be much more proactive in defining standards and laws for what must be collected,” he said. “Otherwise, the automakers and data aggregators will privatize the data and we’ll all be left with an expensive mess to clean up in the future.”

ON THE ROAD

While researchers like Pack sort out the data piece of the puzzle, others are working to roll out a connected vehicle infrastructure on the ground. One of them is Henry Liu, a U-M civil engineering professor who is testing a network of sensors and radio transmitters at UMTRI.

“Right now, we’re working on two things: How can we use connected infrastructure and vehicles to manage traffic better and how can we use information to shorten the development time for driverless cars,” he said.

In the system that’s currently being tested, cars equipped with specialized transmitters send their location, speed and direction of travel 10 times per second to receivers mounted on traffic signals. In return, the traffic signals broadcast their status back to the cars, where it’s displayed to drivers either on a built-in dashboard display or on a standalone device like an iPad. Drivers can use the information to moderate their speed and avoid red lights, and the system can also turn a red light green when a connected vehicle rolls up to an empty intersection.

For now, the on-the-ground applications of the current system are modest, but Liu says it’s generating valuable data that he and other researchers are using to test the components of the system itself and gain insight about how best to roll it out on a large scale.

One key test bed for the technology is Mcity, a 32-acre self-contained cityscape at U-M that’s the world’s largest dedicated testing facility for driverless cars, as well as the data and infrastructure technologies that will go with them. Officially opened in July of this year, Mcity is a public-private partnership that boasts nearly five lane-miles of roads, reconfigurable building facades, traffic signals, roundabouts, fire hydrants, even artificial pedestrians to dart out in front of driverless vehicles. The facility is a major testing site for other technologies as well, including the autonomous car navigation algorithms that Ford Motor Company is developing as part of a multi-million dollar agreement with U-M.


BETTER THAN THE REAL WORLD

Public roads are often touted as the ultimate proving ground for driverless technology. But in truth, they’re only part of the equation. That’s because many of the hurdles that are likely to trip up driverless technology just don’t happen often enough to test, or are too dangerous to test on public roads. What if a pedestrian darts in front of a driverless car? How will a driverless vehicle respond to a police officer directing traffic? What if a stretch of road has no pavement markings? What if another car runs a red light?

The need to make sure that driverless cars and connected vehicle infrastructure work reliably in these scenarios – and countless others like them – is part of what led to the creation of Mcity, the world’s largest testing facility for connected and autonomous vehicle technology.

Located on 32 acres on U-M’s North Campus and funded through a partnership that spans U-M, industry and government, the facility includes nearly five lane-miles of roads. With customizable traffic signals, simulated buildings, benches, roundabouts, street lights, construction barriers and even mechanical pedestrians, it offers engineers everything they need to test driverless technology in ways that would be impossible on public roads.

“In the real world, I can’t ask a pedestrian to walk in front of a car over and over again,” explains Jim Sayer, a research scientist at UMTRI. “But at Mcity, we can run any scenario as many times as we need to. We can control pavement markings, signal timing, emergency vehicles. It enables us to determine very precisely that we’re getting results we can rely on.”

Infrastructure ideas that are in testing right now include technology to control traffic signal timing dynamically in real time, reroute individual vehicles to avoid traffic tie-ups and give signal priority to public transit vehicles. Ford Motor Company is also using Mcity to test its new driverless vehicle navigation systems, along with other manufacturers and suppliers.


In addition to the connected infrastructure testing at Mcity, U-M has turned entire swaths of Ann Arbor into a real-word testing ground for connected infrastructure, outfitting 27 square miles of the city with connected infrastructure and recruiting 9,000 volunteers to connect to the system.

It’s still early days for connected vehicle technology, and Liu says it will likely be decades before it’s used in large numbers of driverless vehicles. But it’s clear that the same data that now alerts drivers will eventually be used to control the car directly.

“When we have a system that can use data to alert drivers, the next logical step is to begin removing the driver from the equation. This will start in the near future with assist functions – for example, your car could automatically brake to avoid a dangerous situation. It’s part of a longer evolution toward autonomous vehicles. And it’s only a matter of time.”

BUCKLE UP

Opinions vary widely on when large numbers of driverless vehicles will hit the streets. But most experts agree on one thing: Driverless is coming. And it’s going to change everything. The goal of safe, commercially viable driverless technology seems closer than ever.

But is the adoption of driverless technology the end of the story or the beginning? Many transportation experts see it as just another piece of a still-evolving, 21st-century transportation puzzle, one that includes not just new ways to get around, but a radical rethinking of what we put into transportation, what we get out of it and how we want it to fit into our lives.

In that sense, driverless technology is more than just a new way to schlep your kid to soccer practice. It’s a catalyst for change. And it’s already sparking conversations and raising questions in a way that oil embargoes, the electric car, light rail and countless other Next Big Things all failed to do. Finding answers won’t be quick, or easy. But it could be our biggest opportunity to rethink transportation in 100 years. And if we want to keep up with the technology, we’d better get rolling.

Media Contact

Gabe Cherry

Strategic Content & Magazine Editor