I’m driving on the Brooklyn-Queens Expressway on my way to my parents’ house when I finally decide to let go of the steering wheel.
As I release my grip, the car veers closer toward the adjacent lane. My palms begin to sweat and pupils dilate. Just as I’m about to careen into the SUV next to me, the car takes control and gently guides me back into the lane.
A few days later, I’m stuck in typical rush hour traffic on Manhattan’s West Side Highway, awash in a sea of lights, blaring horns, and creative insults slung from open windows. I tap a button on my car’s steering wheel and let go. I slide my feet away from the gas and brake pedals and slump back in my seat.
My car takes over. It keeps me centered in my lane, speeds up when it’s time to move and brakes when the car in front of me slows down. My car is driving itself. And it’s both awesome and slightly terrifying.
No, this isn’t some dispatch from the far-flung future; this happened last week. The car in question is Volvo’s 2016 XC90 with Pilot Assist, one of the most advanced vehicles you can buy.
Is the XC90 truly a self-driving vehicle? Not really. Sure it accelerates, brakes, and steers without any help from me, but its autonomous technology is limited to a top speed of 30 miles per hour, it can’t change lanes, and it’s only meant for use in highway traffic.
Despite those limitations, Volvo’s SUV is an impressive piece of technology, and proof that fully autonomous vehicles have come a whole lot closer to becoming a reality in a very short time.
So when can you expect to jump into your car each morning and catch some extra z’s as it drives you to the office? Perhaps sooner than you think.
The car of the future, today
OK, so you can’t head down to Crazy Larry’s Car Dealership and buy a self-driving car just yet. But if you haven’t purchased a new vehicle in the past few years, it might surprise you to learn that automakers are already equipping their cars with near-autonomous driving features that lay the groundwork for a fully hands-free automotive future.
Take the Volvo XC90, for example. The SUV’s Pilot Assist uses a series of cameras and radar to track the vehicle in front of you and follow lane markings. The car can then basically drive itself up to speeds of 30 mph. But it won’t work unless there’s a car in front of you for the vehicle to track.
Other automakers offer lane-keeping assist, which ensures you stay in your lane, as well as adaptive cruise control, which uses radar to detect when you get too close to the car in front of you and automatically slows the car down to keep you at a safe distance.
But that’s today. In its upcoming S90 sedan, due out this summer, Volvo is increasing the top speed of its Pilot Assist feature and eliminating the need for it to track a vehicle in front of you.
“[The S90] will steer itself up to 80 mph … and accelerate and decelerate to a complete stop,” says Jim Nichols, Volvo’s technology and communications manager.
As an example of how pervasive this kind of semiautonomous technology has become, Hyundai recently announced that its 2016 Elantra is the first car costing less than $20,000 to offer adaptive cruise control as an option.
“That’s the democratization of these types of technologies,” says Cason Grover, senior group manager with Hyundai’s vehicle technology planning. “We think it’s important to bring [this technology] into other segments.”
Tesla, meanwhile, offers an Autopilot system that allows its cars to navigate on highways in much the same way as Volvo’s system, but with the added benefit of being able to change lanes when you tap the turn signal.
Today’s cars can even do things like detect people in crosswalks, bicyclists, and large animals, then help you to avoid them all.
Why are car-makers going to such efforts to build cars that drive themselves? Because it makes them safer.
According to a study by the International Institute for Highway Safety (IIHS), vehicles equipped with automatic braking systems cut rear-end crashes by 40 percent. The study suggests that if all vehicles had the same types of systems installed, they could have prevented about 700,000 accidents.
But getting vehicles to the point that they can prevent 100 percent of accidents is going to take a lot more work.
Driving to the future
On a cold day this past December, a pack of Ford Fusions drove themselves over snow-covered roads and into history. The Fusions are part of Ford’s autonomous vehicle fleet, which the automaker is testing at the University of Michigan’s 32-acre Mcity self-driving research course.
And while driving on snow may not seem like a big deal — at least to people outside of Florida — it was actually a major step forward in Ford’s quest to create a fully autonomous car.
Ford, like Google and like other automakers, uses a system called LiDAR (light detection and radar) and cameras to help its vehicles “see” what’s around them. But when the road is covered in a sheet of snow, the cars can’t see where they are. To address this, Ford loaded its cars with high-resolution maps that include what’s on and above the road.
“[The maps] contain the stuff that’s normally seen on maps, but it also contains all of the data that the LiDARS would see,” explained Greg Stevens, Ford’s global manager of automated driving.
“By comparing what our LiDARS see at the current moment versus what they should see on the map, we can figure out exactly where we are on the map. And that’s one of the key things that allowed us to drive in the snow.”
LiDAR isn’t the only technology companies are looking toward to ensure that their vehicles will be able to drive themselves.
Tesla founder Elon Musk has previously stated that he doesn’t believe LiDAR is necessary for self-driving cars, and that optical sensors and radar will work better. LiDAR’s biggest drawback is that it sticks out from the top of a vehicle, creating drag.
(It also looks like a soda can that’s been glued to a car’s roof. So it’s not exactly attractive.)
In addition to LiDAR, a number of automakers are also looking toward a technology called vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. Both V2V and V2I will allow automobiles to communicate with each other and with connected infrastructure systems, such as road signs and traffic lights.
V2V- and V2I-connected vehicles will use wireless radio signals to tell each other where they are on the road and how fast they are moving, as well as let traffic lights warn cars when they’re about to change. The information can then be sent to the driver or used by vehicles to stay clear of each other.
In 2014, the National Highway Transportation Safety Administration announced its support for V2V and V2I communications, saying they could help reduce traffic accidents.
V2V communication might not be that far off, either. In 2014, GM CEO Mary Barra said the company’s 2017 model year vehicles would be available with V2V technology.
For all that’s already been accomplished toward making self-driving cars a reality, there is still more work to be done. Most of the technology being worked on today, though impressive, still isn’t ready for prime time. The fact that driving in the snow is considered a breakthrough should be proof enough of that.
But there is an even more fundamental hurdle autonomous vehicles have yet to overcome: driving on surface streets. See, driving on highways is a relatively predictable experience. Every car is (hopefully) moving in the same direction, there are no traffic lights, lane markings are clear, and you never have to make sharp turns.
That’s why Tesla’s Autopilot-equipped vehicles are able to operate autonomously during highway driving. (It’s worth noting, however, that Tesla advises drivers to keep their hands on the wheel while using Autopilot.)
Driving on surface roads is far more complicated. Even with V2V and V2I systems in place, cars will have to contend with traffic lights and road signs; there are also pedestrians, bicyclists, double-parked cars, and a litany of other obstacles.
Highly autonomous is relatively easy to have in a freeway driving experience, because it’s fixed. “Children don’t run into the street on highways,” said Hundai’s Grover.
“You’re not going to encounter a red light on a freeway,” he says. “Once you get into signal status, signal communication, and communication with signs, you’re working with thousands and thousands of communities and the states.”
But the most significant issue that needs to be worked out is how self-driving cars will be regulated. Right now only a handful of states allow for autonomous vehicle testing on public roads. Those include California, Nevada, Florida, and Michigan, and Washington, D.C. But even those states’ rules don’t have much uniformity, and that’s a problem.
California, for example, has proposed a law that would make it illegal for a self-driving car to operate without a driver. That’s not the case in other states.
In fact, California’s proposal earned a stern rebuke from Chris Urmson, Google’s director of self-driving cars.
“Safety is our highest priority and primary motivator as we do this,”
Urmson said in a statement replying to the proposal in December. “We’re gravely disappointed that California is already writing a ceiling on the potential for fully self-driving cars to help all of us who live here.”
As it stands, there is no national consensus indicating who is liable when a self-driving car is involved in an accident, or how insurance companies will handle this. And that’s one of the most basic questions.
Fortunately, President Obama announced in January that his 2017 budget proposal would include a $4 billion line item dedicated to speeding up the development of self-driving cars over the next 10 years.
That cash will go toward pilot programs that will allow automakers to operate self-driving vehicles in the real world. The move would also help establish a kind of framework for laws related to autonomous cars.
When you’ll buy them
Naturally, that raises the question: When will you be able to buy a self-driving car? The answer largely depends on whom you’re asking, as well as how you define “self-driving car.”
According to the Society of Automotive Engineers (SAE) International, autonomous vehicles are rated on a scale from Level 0 to Level 5. Level 0 vehicles don’t use any kind of autonomous technology. Levels 1 and 2 vehicles use semiautonomous technology like the Tesla Model S’s Autopilot and the Volvo S90′s Pilot Assist.
Level 3 vehicles are largely autonomous but require drivers to be able to take over when their car alerts them of a hazard. Level 4 vehicles are completely autonomous and will alert drivers when there are hazards, but doesn’t require them to take over, while Level 5 vehicles are fully autonomous.
Ford CEO Mark Fields recently stated that the company expects to have fully autonomous vehicles on the road by 2020.
Hyundai says it’s likely to have more semiautonomous highway driving features by 2020, with full autonomy coming by 2030. Honda likewise says it will have autonomous highway-driving features available by 2020. Fully autonomous vehicles will come sometime after that. And by 2050, Honda believes, autonomous vehicles will be so prevalent that we’ll be living in a zero-collision society.
Toyota also believes its vehicles will be able to drive themselves under certain conditions by 2020, though full autonomy is still decades out. Toyota is currently testing autonomous highway driving with its Highway Teammate Lexus concept.
“The Highway Teammate Lexus will have high-level autonomous driving on the market by 2020,” explains John Hanson, national manager of environmental, safety and quality communications at Toyota.
“It won’t be fully autonomous unless you’re in certain situations. The key to this is how cars can drive safely for themselves where it’s an environment of other cars without pedestrians or bicyclists.”
Audi is following a similar timeline for its self-driving vehicles, saying that the technology is about 20 to 30 years away from becoming a reality.
“You need to have the algorithms to understand different laws, how people drive, different customs, and so on,” says Brad Stertz, senior manager of corporate communications for Audi.
Jaguar-Land Rover is far more bullish on autonomous vehicles, saying that we’re less than 10 years away from a self-driving society. Google is of a similar mind. Tesla has an even more ambitious timeline, with CEO Musk stating that cars will be fully autonomous within the next two to three years.
In fact, Musk believes you’ll be able to summon a vehicle parked in Los Angeles from your smartphone while in New York, and it will be able to drive cross-country to you.
Volvo, meanwhile, believes it will be the first automaker to put fully autonomous vehicles on the road starting next year.
That program will put 100 autonomous vehicles in the hands of consumers in Volvo’s home country of Sweden. Those vehicles will, however, be fully autonomous only within an area Volvo has already mapped for use by the cars.
So which company should you believe? Well, despite Tesla and Jaguar’s predictions, getting consumer-grade self-driving cars on the road in less than 10 years sounds like a bit of a stretch.
Still, the technology needed to power these vehicles is moving incredibly fast, with new developments being reached and barriers falling at a record pace.
“The thing that’s interesting to me is looking back and seeing how fast things have changed,” said Ford’s Stevens. “Two to three years ago, there were still skeptics, but some people were starting to say the engineers are making process. And now it’s becoming a reality.”
In other words, our self-driving future could be here sooner than we think. Buckle up, it’s going to be a heck of a ride.