Motoring along New York’s Harlem River Drive the other day, I found myself in a sea of traffic. There were honking taxis on one side and a large truck fast approaching from behind. Add to that buckling pavement, which made it feel like I was on a roller coaster.
Normally, a bumper-to-bumper drive like this would have my heart racing. But for this drive, I was behind the wheel of a Mercedes-Benz S550 Coupe, which happens to have one of the most advanced suites of semi-autonomous technologies on the market.
You heard me right: a car you can buy today at the auto dealer has a whole lot of self-driving car technology already in it. In this Mercedes, I could sit back in its supple leather seats and let the car do the heavy lifting. Thanks to a radar-based system that the automaker calls Distronic Plus with Steering Assist, the car could—on its own—slow and brake as the traffic came to a halt and then sped up again. (The system works up to 120 miles an hour.) In this traffic, I was happy to let it do so.
“You should feel that taking control,” said Kaitlyn Aurilio, an assistant product manager at Mercedes-Benz, as she watched my hands from the passenger seat. “As we get closer to other cars, it will automatically brake. Gently take your hands off the wheel. You should feel the steering input. You should feel it pulling you in.”
It was hard not to be nervous driving a $120,000 vehicle in New York City traffic. But the two-door, anthracite-blue coupe was just getting started. As we headed off the drive and cut through Manhattan, I had the chance to try out several features that take advantage of the 12 ultrasonic sensors—several radars of varying lengths as well more half-dozen cameras that offer a 360-degree view around the car—embedded in the vehicle.
When I began to stray from my lane, the car’s Active Lane Keeping Assist pulled me back. When I attempted to change into a lane occupied by a fast-charging minivan, its Active Blind Spot Assist beeped at me. The vehicle also has a feature called Attention Assist, which Mercedes says uses 75 parameters to determine if a driver is getting drowsy and alert them with a beep and a message: “Time to break.” At night, a feature called Night View Assist Plus uses infrared cameras to detect animals or pedestrians in the car’s path and warn them—with a flash of its headlamps—that there’s a two-ton piece of steel and aluminum headed their way.
I made my way onto Riverside Drive along Manhattan’s western edge. It was time to park. Rather than slowing to a crawl and craning my neck only to endure the honking of angry motorists behind me, I engaged the car’s Active Parking Assist. It scanned the rows of cars along the street for a spot of the right proportions and alerted me with a arrow in the instrument cluster when it found one. I threw the car into reverse, pushed an “OK” button and the car did the rest—no New York “tap” necessary.
“Once drivers have experienced the active braking as part of the Active Blind Spot Assist or the lane keeping assist, they can’t go back,” Aurilio said. “They feel that sense of security and safety when the vehicle steps in where the driver fails or needs that assistance.”
A revolution underway
There is great anticipation—and a not-insignificant dose of trepidation—for a future world in which cars simply drive themselves. Google, through its experiments in California and other western U.S. states, has sparked a broad fascination with the autonomous vehicle that hasn’t been seen in decades. Some applaud a vehicle that is demonstrably safer than a human-guided one. Others worry about the issues—legality, liability, experiential, cultural—that such vehicles raise.
In truth, many people don’t realize that self-driving technologies have already made their way into today’s conventional cars. Today’s vehicles—from suburban minivans to sleek coupes like the Mercedes—are chockablock with sensors working on our behalf. Driven by a desire for greater comfort and safety, automakers from Detroit, Michigan to Wolfsburg, Germany have embarked on a technological revolution that is, with each additional sensor and algorithm, handing greater control of the driving to the car. In other words, the autonomous vehicle revolution is already upon us—we just don’t realize it.
“The automakers are already deploying a lot of advanced technology,” says Peter Sweatman, the director of the University of Michigan Transportation Research Institute. “Things like lane-departure warning systems, forward-collisions warning systems, and all these side-object detection systems are really rolling out through products for all auto manufacturers. There has been a tremendous technology revolution in the past 10 years. Automakers view this move toward connectivity and automation as really another stage of that kind of technological development.”
The technology couldn’t come at a better time. According to the report “Self-driving cars: The next revolution” by the consulting firm KPMG, there were 6 million crashes in the United States in 2010 resulting in nearly 33,000 traffic deaths. Ninety-three percent of those were caused by human error. Crashes sent 2.3 million people to the emergency room and cost some $300 billion every year. Traffic collisions in the United States are the leading cause of death for those between the ages of 4 and 34.
There are other benefits of these autonomous technologies. The average commuter spends 250 hours a year behind the wheel of a vehicle, resulting in time lost for other activities and a dent in productivity, according to the KPMG report. The search for parking, especially in dense urban areas, is a substantial waster of time and energy: According to the MIT Media Lab, city drivers use 40 percent of their fuel searching for a parking spot.
“[Autonomous driving systems] are taking a load off the driver on long, monotonous drives. That’s a convenience factor,” says David Alexander, a senior researcher with Navigant Research’s Smart Transportation Group. “It is also a safety factor. If you’re cruising along on the freeway at 70 miles per hour and you’ve been doing it for a few hours and suddenly something moves in front of you, the system will automatically slow you down. In the past, if you are on cruise control, the car would plow on at 70 miles per hour regardless of what was in front of it.”
A trickle-down effect
Alexander says such systems were first introduced a few years ago in “very high-end luxury class vehicles” such as BMW’s 7 Series or the Mercedes-Benz S-Class. Today, these technologies are available in moderately priced models such as Ford’s Fusion. “These core systems are now becoming very widespread, very available across a full range of vehicles,” he says. “At the high end, they must look for things that will differentiate them from that. That’s why they are moving on to semi-autonomous things where you combine two or more of these systems.”
Which is why an economy model today likely has electronic stability control (which detects loss of traction) or automatic braking (which senses an imminent collision with a vehicle) but not what the industry calls “Level 2” systems, such as adaptive cruise control paired with lane centering, that are found on today’s luxury models.
Volvo’s XC90 T6 sport-utility vehicle (which sells for approx. $50,000) is one such high-end model. When it begins shipping next year, the vehicle will come with a veritable laundry list of Level 2 features. It will have the ability to detect and avoid collisions with other vehicles, pedestrians, and cyclists; it will be able to park itself in perpendicular spots; and, unique to Volvo, it will brake when making a left turn if it senses an oncoming car.
Tesla’s Model S (approx. $70,000), which is available today, comes with long-range radar, image recognition (so the vehicle can “see” things like stop signs and pedestrians), and a 360-degree ultrasonic sonar. Meanwhile Nissan in July announced that it plans to equip various models over the next four years with Level 2 capabilities such as automatic lane control, highway traffic management, and automated parking.
“This will be followed in 2018 by the introduction of multiple-lane controls, allowing cars to autonomously negotiate hazards and change lanes,” said Carlos Ghosn, president and CEO of Nissan Motors, to reporters in July. “And before the end of the decade, we will introduce ‘intersection autonomy,’ enabling vehicles to negotiate city cross-roads without driver intervention.”
General Motors (GM) plans to offer a Level 2 feature known as Super Cruise on its 2017 models that includes hands-off lane following, braking, and speed control in certain highway conditions.
“A tide of innovation has invigorated the global auto industry, and we are taking these giant leaps forward to remain a leader of new technology,” said General Motors CEO Mary Barra last month at the Intelligent Transport System World Congress in Detroit. “We are not doing this for the sake of the technology itself. We’re doing it because it’s what customers around the world want. Through technology and innovation, we will make driving safer.”
The next tech wave
GM’s Super Cruise will be one of the first Level 2 features to use vehicle-to-vehicle, or V2V, technology, which allows cars to directly talk to one another. GM and its peers say the technology will help reduce traffic collisions and congestion by transmitting critical information—such as the location, speed, and direction of an approaching vehicle—to others around it.
The U.S. Department of Transportation seems to agree. It announced earlier this year that it would be begin taking steps to enable V2V technology for light vehicles. Anthony Foxx, the department’s secretary, said the innovation represented the “next generation of auto safety improvements, building on the life-saving achievements we’ve already seen with safety belts and air bags.” The DOT projects that the system could prevent 592,000 crashes each year and save 1,083 lives at an estimated cost of less than $350 per vehicle by 2020.
The U.S. Department of Transportation has been working on V2V technology for more than a decade and is involved in the largest deployment of it to date at the University of Michigan. The university plans to triple the number vehicles that it has equipped with V2V technology, to 9,000, as part of a test that will be conducted over the next two years. The evaluation is led by the university’s new Michigan Mobility Transformation Center, run in partnership with 13 companies (including GM and Ford) as well as the Michigan Department of Transportation.
“You can think of it as world’s best sensor,” Sweatman says. “This is a sensor that can see around corners and see inside other vehicles and receive information about the weight of another vehicle and what other vehicles are doing in terms of braking.”
The university is also expanding the use of what is called vehicle-to-infrastructure, or V2I, technology to 20,000 vehicles across southeast Michigan. The system—a collection of radios, cameras, and sensors deployed on more than 100 miles of roadway—relays relevant information about the road ahead to connected vehicles, such as a traffic accident or slick roadways. The technology, which could begin large-scale deployment in the next five years, will help improve today’s Level 2 features and bring automation another step closer to reality, Sweatman says. “We are getting to the point where we won’t have crashes from a technological point of view,” he says.
Alexander argues that V2V and V2I technologies require heavy adoption to make an impact. “You can only tell if a car fitted with the same V2V system as you’ve got is coming; you can’t tell if anyone is coming,” he says. “And there is always the question that once we get all the vehicles [equipped], then you need all the motorcycles, all the bicycles, and the pedestrians to be [included].”
Some automakers are looking beyond Level 2 technologies entirely. Volkswagen’s Audi subsidiary is gambling that Level 3, or limited self-driving automation, technologies are the way forward. The piloted driving prototype of its A7 luxury sedan, which features 23 sensors and cameras and a laser scanner in the front grille, carries a system called Traffic Jam Pilot that allows for hands-free driving on the highway up to 40 m.p.h.
The test vehicle won’t be ready for production for another 4.5 years, the automaker says, but that hasn’t stopped it from taking the experimental A7 for a spin on the Lee Roy Selmon Expressway near Tampa, Florida in July. More recently Audi put the vehicle through its paces at Germany’s Hockenheimring race track at speed—sans driver. “We want to be the first one to get into Level 3,” says Thomas Mueller, Audi’s head of development for braking, steering, and driver assistance systems. “We believe this is where you generate something different for customers. They will feel, when driving with Traffic Jam Pilot, ‘Wow, that was a great experience. I didn’t have to do anything. I could Skype with my wife and I was relaxed in the end.’ He will build up trust because he won’t have to take over.”
Marcus Rothoff, Volvo’s autonomous driving program director, says Volvo was equally motivated by customer complaints about worsening traffic jams and a desire that they “do something else with their time.” “Many incidents we have in urban areas are due to distractions and nearly all accidents have some kind of human error involved,” he says. “When we take away the possibility of human error in the driving, we, of course, can make a much safer car.”
It’s one thing to road-test a prototype; it’s quite another to bring an autonomous car to market. Automakers interviewed for this article acknowledged that there remain many technical challenges for the semi- and fully autonomous vehicle, none more important than developing a network of automotive sensors that can match, if not surpass, the ability of the human eye to quickly identify any number of hazards.
Those limits were on display in June when Carnegie Mellon University was forced to abort a test around Washington D.C. of its autonomous Cadillac SRX for 20 members of Congress. The system seemed to have a hard time adjusting to, on one day, the emergence of a temporary construction site and, on another day, cyclists passing near the car. “The urban setting turned out to be a lot more complex,” said Raj Rajkumar, co-director of the GM-CMU Autonomous Driving Lab in Pittsburgh. “Truth be told, we actually had to take control a couple of times in these rides. We knew the vehicle might not do the right thing.”
Ditto the driver of a semi-autonomous vehicle, who may flout road laws to abuse its technical capabilities. “I don’t want to have people . . . sitting in the passenger seat and no one is sitting on the driver seat,” Mueller says. “I don’t want to see that. I want to make driving safer, not unsafe.”
Some of those protections will come through legislation. Most states don’t yet allow testing of autonomous cars, let alone regular operation. Some will come through additional technologies, such as back-up braking systems that bring the car to a safe stop when the driver doesn’t respond. Some will come through focused investment to address America’s overtaxed and often poorly maintained road networks, which can neutralize even the most advanced of today’s driving technologies.
That was apparent several times during my ride around New York City in the S550. Often, my attempts to turn on the lane-keeping system were thwarted by the absence of visible lane markers on the road. Clearly frustrated, Aurilio promised the car’s system would prove more consistent across the river in suburban New Jersey, where roads were better maintained.
No matter. Despite a few speed bumps, it’s clear that the autonomous vehicle revolution underway will continue to accelerate. “Driving is a complicated thing that we do, but it’s not brain surgery, so to speak,” says John Capp, director of GM’s global vehicle safety strategy and vehicle programs. “It absolutely makes sense that computers and computer driven vehicles will be capable of handling the driving task and the situations we encounter.”