Autonomous vehicles coming of age
Imagine a nearly perfect fall day, barely a cloud in the sky, sun shining bright, cutting an amber glare across your windshield. You're playing hooky from the office to spend some quality time with your tween-age daughter before she grows up and loses interest in such things.
Strapped into the front passenger seat, she gleefully hums her favorite Taylor Swift song while enjoying a bag of snacks as you navigate your upwardly mobile sedan toward the downtown shopping district. Then suddenly...bam! You're broadsided making a left turn onto the main thoroughfare just down the block from your destination. The driver wasn't paying attention and ran the red light at approximately 40 mph, 5 mph above the posted speed limit. Upon impact, your car spins around like a top several times before it flips over and comes to rest on its roof.
Fortunately, your car is equipped with front- and side-impact air bags on both the driver and passenger sides. Seat belts keep you and your daughter in place, even while hanging upside down.
Later when the police summon the wrecking crew to haul the carcass of your mangled vehicle off the roadway—well after you are released by EMTs with only minor bumps, bruises, and scarred memories—they'll relay the good news:
For the past 50 or so years, the purpose of automotive safety has focused on mitigating the effects of a car crash; limiting the physical damage caused by metal-to-metal mishaps. But that's changing. The next generation of automotive safety devices and driver assist technologies will focus on both the physical and psychological aftereffects of a collision, by removing the possibility of even getting into one. The breakthroughs that might save lives today are the building blocks in the seemingly far-off technology of tomorrow: driverless cars.
Computers nearly mature enough to drive
The technology that will prevent highway horror stories is evolving at a blistering pace, and by and large we have the quest for autonomy to thank. "When you look at where autonomous driving is going, it's the fusion of an awful lot of data," says Kent Helfrich, our vice president and chief technical officer for the automotive division of global design, engineering, and manufacturing. "So, there are cameras, radars, adaptive cruise control, and lane departure warnings—these are things that are already in the marketplace."
Automakers are beginning to think even more deeply about their direction, adds Chris Ricci, our deputy general counsel. "Traditionally, the auto industry has been a mechanically oriented business. However, it's quickly becoming more electronic, more programming oriented." And this paradigm shift has caused car companies to react at a rate never seen before.
For instance, in just one model year, the number of new vehicles on the road that offer adaptive cruise control, blind-spot monitoring, mitigated braking, or other advanced driver assist systems (ADAS), the so-called building blocks of autonomous driving, has grown from the minority into the majority.
Take lane departure warning, an ADAS feature that alerts a driver when his vehicle is drifting outside the lane, and Lane Departure Prevention, which actually helps steer that vehicle back between the white lines. "The number of vehicles in which they are available has jumped from 44% in model year 2014 to more than 63% in model year 2015," says Karl Brauer, senior director of insights and industry consultant for Kelley Blue Book, a vehicle valuation and information source. "This much growth in one year is unheard-of in the auto industry."
Over the course of five years, such growth is even more staggering, especially in terms of popularity with consumers. Revenue earned from the top three ADAS technologies (adaptive cruise control, lane departure warning, and blind-spot monitoring) has jumped from $1.05 billion in 2010 to more than $6 billion this year, according to technology market analysts at IHS Technology.
"ADAS technologies are becoming the next competitive area, a place for one automaker to differentiate its vehicles from another's," explains KBB's Brauer. Even though the market is still in its infancy, the systems are already well executed and well-integrated, even among affordable vehicles. Only a few years ago, these active driver assists were offered only on higher-end luxury vehicles.
Today, active safety systems are well on their way to becoming democratized. You can buy a 2015 Honda CR-V—an excellent but far-from-upscale compact SUV—with a slew of automated systems. Opt for Honda's Sensing(TM) package and you get technology that can regulate the car's speed depending on the vehicle in front (known as adaptive cruise control), alert you to a motorcycle or car in your passenger-side blind spot (LaneWatch(TM)), or even avoid an accident altogether (collision mitigated braking system).
We need help at the wheel
Fact is, humans are terrible drivers (see: NHTSA statistics on driver behavior). "We're not anatomically designed to handle a vehicle moving at 70 mph and perceive and react to stimulus," explains Bryan Reimer, associate director of the New England University Transportation Center and research scientist at MIT's AgeLab. He leads a team of researchers and students who study how automation can help drivers avoid distractions in increasingly complex driving environments. "We're daytime hunters by nature. That means our eyes are tuned to see pretty well during the day, but not so much at night—especially as we age. Consequently, our essential systems tend to pick up objects [like pedestrians] at up to running speed, not highway speed."
Sadly, most people think they are good drivers, but their actions behind the wheel tell a different story. In 2011, insurance firm Allstate surveyed 1,000 people about their driving habits. Nearly two-thirds rated themselves as excellent or very good. Yet many copped to unsafe practices that put themselves and others at risk.
At the very least, humans are inconsiderate drivers. We get angry or frustrated by the simplest of offenses, frequently losing our patience and lashing out irrationally at those around us. We thoughtlessly drink and drive. Speed and take needless chances. Send texts and talk on the phone. All while trying to pilot a several-thousand-pound missile traveling at 60 mph.
A computer will never do any of those things. Still, today it's just barely better than humans are behind the wheel. But unlike us, it's getting better—and quickly.
Currently, computer-driven cars can only handle certain situations well. They perform almost perfectly on the highway and in other mostly straight-line driving scenarios. Problems arise when the computer is asked to interpret more unexpected activities of other drivers, or to account for random factors on the road. "We call these scenarios Edge Conditions," says Don Norman, director of the Design Lab at the University of California, San Diego, where he and his team are applying human-centered design to complex social and technological problems, including automated driving.
"Unpredictable variables arise from interacting with these conflicts, the social environments of the city and the streets. We can solve the highway issues somewhat easily—much fewer variables [mostly straight-line driving]. But when you add bicycles, motorcycles, people, pets, kids, vague directions, varying weather conditions, and other outside influences, computers cannot yet interpret all of the data in a suitable time and manner. We'll just have to grow to accept the fact that there will always be some accidents because there are cases where neither computers nor people will be able to react in time, but computers will still be far safer than [humans]."
But the point is not about who is the better driver or who has the potential to be the better driver. It's who we think is the better driver—perception versus reality. To arrive at a future when we can feel confident about taking our hands off the wheel to enjoy the morning paper and a spill-free cup of coffee, or to engage in a videoconference or to check email, while our cars shuttle us to work, school, or our kids' soccer practice, we have to believe that self-driving cars never make mistakes.
That will take a serious amount of trust—in the technology and those who create it. "Are you willing to give your life to a computer?" asks MIT's Reimer. "I am not comfortable doing so at this moment. Computers crash all of the time; Windows does things that I can't predict all of the time." Could this happen in a car? Yes. And Reimer isn't alone in his discomfort; a 2014 online survey by Seapine Software recently found that 88% of Americans are still afraid to ride in an autonomous vehicle.
Consequently, the level of trust needed to make people comfortable behind the wheel of a self-driving car will take some time. But it will come. "Unfortunately, that's probably going to take far longer than is in our best interest," says Reimer.
Benefits on the road to autonomy
While truly hands-free driving for the masses is still some ways off, that doesn't mean the technologies that drive autonomy can't help us today. The consensus among experts is that these smart technologies are already beneficial, helping us to drive safely and more efficiently than we can on our own.
By setting our sights on the loftier goal of autonomous driving, we've actually enabled a kind of rolling revolution, one in which academics, carmakers, technology companies, and governments have developed some pretty amazing technologies that can change people's lives.
First, they will do so by helping us stay alive and more alert while driving. Second, they'll encourage us to be better stewards of the planet. And finally, they'll aid in our productivity (when we hit that ideal state when we don't even have to think about safety and efficiency). In other words, on the road to driverless vehicles, the journey is more important than the destination.
The promise of greater safety is where these building blocks of autonomous driving come into their own. In 2013, 32,719 people died in traffic crashes in the U.S., according to the Department of Transportation's National Highway Traffic Safety Administration (NHTSA), a 25% decline in overall highway deaths since 2004. Ultimately, though, human error is still the cause of most collisions—93%, according to a widely cited and definitive study funded by the department.
How many lives can these technologies really save?
Even fender benders have a significant impact—not so much in terms of body count, but in dollars spent to fix the damage. Traffic incidents in the U.S. cost more than $275 billion in damages in 2010, up from more than $230 billion the decade before. When quality of life is taken into consideration, the total value of societal harm from motor vehicle crashes in 2010 was almost $871 billion. This cost increase has occurred even as accident rates in the U.S. have fallen.
But as driverless cars become more ubiquitous over the next few decades, experts predict that the effects of car crashes will diminish exponentially. Until then, we will have to rely on semiautonomous systems to make a difference, which could eliminate up to 80% of collisions.
Cars need better communication skills for starters
Before machines can even consider rising up, taking control, and eliminating collisions altogether, they need to be properly socialized. In other words, they need to be able to communicate and interact better. Automakers, academics, governments, and technology companies all believe the answer can be found within the Internet of Things.
The connected car is already changing the world of autos. New York-based consulting firm McKinsey & Company expects the number of networked cars to increase by 30% annually until the end of the decade. The result: More than 20% of all cars will be connected to the Internet by 2020, compared to 10% by the end of 2015.
Right now, we are seeing more models from the likes of Ford, General Motors, and other automakers come with fast 4G mobile broadband, which brings increased data processing and access. Today, this connectivity allows our cars to be conduits for digital lifestyle services that we deem essential—things such as streaming audio and video, social networks, and telephony functions. This is why our four-wheeled friends are fast becoming the ultimate mobile device and a major cog in the Internet of Things.
Such connectivity does little to improve safety. But for more than a decade, academics, automakers, government agencies, and technology wags have been working on a wireless communication system that has the potential to deliver a bigger, more immediate impact on road safety—but only when used in conjunction with advanced driver assist technologies. Such impact is even greater than those promised by advanced vehicle automation technologies.
Known as V2X, this communication system is designed to allow cars to network with one another, the infrastructure, and the cloud.
Three overlapping vehicle communication systems
1. Vehicle to Vehicle (V2V) Communications allows cars in the same geographic area (the dedicated short-range communication specification for onboard equipment broadcast is approximately 1,000 meters) to exchange information such as vehicle speed, location, and direction of travel. Each vehicle broadcasts a Basic Safety Message and the other vehicles listen (when not broadcasting their own BSM). Vehicles then take the "listened to" messages and build models of where the other vehicles are. From these models, the safety system then triggers the user interface to inform the driver of vehicles in the area. With this data, vehicles can then determine whether a crash is likely and warn drivers to stop or slow down to avoid contact. In more advanced systems, it can even brake the car automatically if a driver doesn't respond quickly enough.
2. Vehicle to Infrastructure (V2I) Communications lets cars interact with roadside infrastructure, such as traffic lights and work and school zones, to get traffic updates and rerouting alerts.
3. Vehicle to the Cloud (V2C) Communications allows your car to connect with your house, office, and other smart devices. It acts as a digital assistant, gathering information you need to go about your day.
Today, any vehicle-to-cloud connection comes courtesy of your cellular modem from a common carrier like AT&T, Verizon Wireless, China Mobile, etc., and that pipe into the car can connect you to great things. "Once we have enough of the vehicle fleet with these systems and this amount of data being shared, we're going to see improvements in safety, avoidance of accidents, and an overall more efficient transportation system," says Flex's Helfrich. "And that's really what our long-term interest is: enabling safer, more efficient transportation systems through the application of technologies."
The most developed of these three communications technologies is V2V.
The NHTSA and the University of Michigan Transportation Research Institute equipped nearly 3,000 cars with experimental V2V transmitters between 2012 and 2014. After studying the corresponding communication records for those vehicles, researchers concluded that V2V technology has the potential to prevent more than half a million accidents and more than 1,000 fatalities in the United States every year. U.S. Secretary of Transportation Anthony Foxx was even more optimistic in his assessment of the system's potential, saying that V2V could reduce by 70% to 80% the number of crashes involving unimpaired drivers (who are behind half of all crashes; the other half is caused by drivers who are drunk or high on drugs). That means V2V could save approximately 12,000 to 14,000 lives every year in this country alone.
You might question the need for a system that sounds overly complex and expensive when many cars already have instruments that use radar or ultrasound to detect obstacles like pedestrians or other vehicles. But the range of radar or ultrasound sensors is limited to a few car lengths; they cannot see past the nearest obstruction. V2V can—and thus paints a more detailed picture of the environment surrounding the car.
There are three major roadblocks to the implementation of V2X communications and an intelligent transport system, however. One is that automotive manufacturers must first agree on standards. Another is that many concerns over privacy (who owns the data collected and how it will be used) need to be assuaged. And finally, a decision has to be made about how we are going to pay for it all. Until these three issues are addressed, V2X is stuck in the slow lane.
Hackers on the highway
Anything that's "connected" is susceptible to being hacked. Cars are no exception. But hacking an automobile is easier said than done, no matter what some media superstars or elected officials claim. Telematics, the systems that send and receive information internally and externally in connected cars and control components of those cars, are attractive targets. They can control critical systems like braking, steering, and acceleration. Even cars that aren't connected are susceptible to cyber attacks. In August 2013, researchers jacked a laptop into the OBD-II diagnostics port of a Toyota Prius and took over the car's brakes, accelerator, and steering wheel. However, to date, there has only been one successful car hack that did not require direct hardware access to the cars systems.
Researchers from the German Automobile Association created a rogue cell phone network that duped BMW's ConnectedDrive app, sent fake messages to a BMW's SIM card, and allowed them to gain access to a car. Once in control, researchers could lock and unlock the doors, but they were unable to get into any critical systems. Is there a solution to such security breaches? Yes. Software makers are focusing on safeguards that let cars self-adjust if they've been compromised, as well as making sure we can trust and verify the information that is being passed from vehicle to vehicle, vehicle to infrastructure, or vehicle to the cloud.
The failure hype does have a benefit, says John Ellis, the former global technologist at Ford Motor Company who now runs his own boutique tech consultancy firm, Ellis & Associates. "The fear mongering gets people to be diligent about [hacker intrusion] because you don't want unintended consequences," Ellis told PC Magazine contributor Doug Newcomb recently. "But it's nowhere near this cataclysmic event that people keep hearing about, and the car companies are hiring security people and taking this more and more seriously."
Safety aside, the same smart technologies can also improve efficiency, from both a sociological and an environmental perspective. Few people would consciously choose to spend the equivalent of a workweek (38 hours) stuck in traffic each year, but it's a fact for the average American commuter.
So why not let the car drive itself while you do some work, read a book, watch a film, or chat with family and friends? Staring intently at the bumper of the car in front is a waste of time.
While the hit to our personal lives and productivity is significant, traffic is actually more detrimental to our bank accounts and the environment, according to two reports: one in 2012 from the Centre for Economics and Business Research and INRIX Inc., a leading provider of real-time traffic information, transportation analytics, and connected driver services; and another from Texas A&M's Transportation Institute, using INRIX data. Roadway congestion has caused urbanites to travel 5.5 billion hours out of their way and purchase an extra 2.9 billion gallons of fuel.
This adds up to 56 billion unnecessary pounds of carbon dioxide released into the atmosphere and has cost the U.S. economy approximately $124 billion. Without significant action in the next few years, that figure is expected to increase by 50%, to $186 billion, by 2030.
Why is it taking so long?
Achieving true autonomy is not going to be easy. First, we're programmed to presume that autonomous vehicles will function a lot like our other electronic gadgets: pretty good but susceptible to the occasional ghost in the machine. Unfortunately, pretty good won't be good enough, in this case.
Computers must think and react in real time, a term that is incorrectly used more often than not, says Derek Kuhn, vice president of sales for BlackBerry Technology Solutions, which includes QNX Software Systems, makers of one of the most reliable and secure real-time computing platforms for the auto-motive environment: "Most general-purpose operating systems are not real time because, as we've discussed, they can take a variable amount of time to react to input."
This is why your PC or computer takes a pause for the cause now and then. "The true definition of a real-time system is one that is engineered to respond to input immediately and correctly—every time," he says.
In the autonomous space, the definition of real time goes even deeper. Driving aids such as blind-spot monitoring, lane departure warning, adaptive cruise control, as well as autonomous driving systems, need to be able to make life-critical decisions in micro- and milliseconds, not minutes or seconds.
If these systems fail while operating, someone could die. Even so, Kuhn and Andy Gryc, former product marketing manager at QNX, agree that computers are faster than humans in certain situations. "Yes, they can think much faster than a human for certain tasks," Gryc told Fast Company early last year. "The average person probably couldn't multiply two huge numbers. And neither could I. But a computer could." However, humans are better at qualitative analysis, and they can recognize and adapt to complicated patterns like three-dimensional shapes. Both of these are essential behind the wheel. "We can also reason and think on our feet and outside the box," Gryc said. "Computers need to be trained to do those things."
In the future
Eventually, we will live in a driverless world, a future first promised to us at the 1939 World's Fair in New York, in which transportation will be safer, more efficient, and allow commuters and motorists to maximize their journey to work, grandma's house, or school. "Cars will still represent a certain amount of freedom, but not as much as the cell phone or computer," Ricci reports. "You no longer have to get out on the road to go somewhere. You just have to push a button."
Your physical body can be in one place while your mind is in another. But that's not possible until sensor technology and software become more robust and less vulnerable to bad weather, unexpected variables, or complex city driving. There is a lot of money to be made—experts say trillions of dollars—in solving these problems. Why else would corporate juggernauts such as Apple, Google, and Tesla invest so heavily in the technologies associated with driverless cars, connectivity, and ADAS?
The mysterious iCar, is a good example. According to media reports, Apple CEO Tim Cook has been putting together an all-star team of industrial designers and lithium-ion-battery experts to build an electric vehicle and examine the in-car environment of the future. Apple has refused to confirm or deny this, but evidence is mounting daily. Says Ellis of the consultancy Ellis & Associates: "No one assembles such an established and expensive group of experts if they aren't dead serious about the mission."
But that mission might not be to build an autonomous or electric vehicle for mass consumption. Instead, Ellis suggests that Apple might be building a vehicle to learn all it can about EVs and the automotive environment and systems in general, so that it can reinvent the wheel, so to speak, without having to become an automaker. But it isn't the bells and whistles or the money to be made from developing driverless cars that matter. It's the experiences that all of these technologies enable. It's about your children never having to worry about driving along and getting T-boned by some distracted driver. Thanks to technologies like V2V, cross-traffic alert, and other ADAS features, they probably never will.