Self-Driving Cars: The Road Ahead

Traffic congestion along busy stretches of roads and highways can make commutes a dreaded daily routine for most California motorists. In fact, after Washington, D.C., Los Angeles and San Francisco are the nation’s second and third most gridlocked cities according to a new report by the Texas A&M Transportation Institute.

But California is also home to Silicon Valley, where major car companies, including Ford, Mercedes and Audi, as well as tech titans such as Google, are engineering a new generation of vehicles: self-driving cars.

Experts say vehicles that drive themselves would not only ease congestion, but also boost safety and allow motorists to work or play as their cars take control of the wheel.

“If self-driving cars are really safe, then crashes, which are a huge source of unexpected congestion, can be eliminated,” said Nidhi Kalra, a robotics scientist who co-authored a paper on self-driving cars for the RAND Corporation in 2014.

Since more than 90 percent of the more than 32,000 motor vehicle deaths in the United States each year are due to human error, such as drunk driving or texting behind the wheel, automating some or all of the driving tasks should substantially reduce fatalities, proponents of self-driving vehicles argue.

And the cars are further along than many people realize. In January, an Audi A7 drove 550 miles from Silicon Valley to Las Vegas with company officials and journalists on board, with the car driving itself at speeds of up to 70 mph. In April, an Audi Q5 modified by Delphi, drove from San Francisco to New York over nine days in autonomous mode 99 percent of the time.

Still, it may be a decade or more before fully self-driving, or autonomous, vehicles are for sale to consumers, given the many technological hurdles that have to be overcome before they can handle the numerous complex driving situations and environments human motorists encounter daily.

“How well can the vehicle perceive all of the hazards and the objects in the environment that are not dangerous?” said Steven Shladover, a transportation engineer at the University of California Berkeley.

Using the example of a paper bag or a newspaper, he pointed out that the sophisticated sensors on a self-driving car may detect these obstacles but fail to detect potentially dangerous obstacles such as a pothole, which he said is “really hard to see by almost any sensor technology.”

Although challenges still remain, Silicon Valley is fast-becoming a 21^st-century Detroit.  And it may not even be a traditional auto maker that is the first to cross the finish line with a consumer-ready self-driving vehicle.

In June, Google began rolling out its latest self-driving prototype, which tops out at 25 mph, on the streets of Mountain View near its headquarters. The company has logged roughly a million miles since it began testing its prototypes in 2009, and its co-founder, Sergey Brin, has said the cars could be available to the public as soon as 2017.

In addition to Google, there are eight other companies in California that currently hold permits issued by the state to test dozens of autonomous vehicle prototypes on public roads.

Experts say that cars will continue to evolve, first from features that have been common for years like cruise control, through technologies available in new cars, like self-parking, to new models in the coming years that will change lanes by themselves, and may eventually be driverless, a trend that could change everything from the taxi industry to the way cities and suburbs are designed.

But first, concerns around the technology and liability have to be overcome. Lawyers and lawmakers, for example, are working out who would be at fault if a self-driving car crashes and injures someone.

Increasingly, even conventional cars are becoming computers on wheels, with WiFi and Bluetooth-enabled communication and infotainment features that can make them vulnerable to being hacked. In July, Fiat Chrysler recalled more than a million of its cars after two experts demonstrated a security flaw they exploited to wirelessly hack into a Jeep Cherokee as it drove down a freeway.

Myles Kitchen, an automotive engineer in Santa Cruz who works as a consultant for auto companies and start-ups, said hackers can control a car’s engine, throttle and braking if they can gain access to a car’s Controller Area Network bus. The device functions like a conductor to relay data messages between the many electronic modules in a car that control everything from power steering to low fuel alerts.

But as critical as the computer centers are for the operation of a car, they are vulnerable to being hacked because they lack firewall protection commonly found in home computers.

“When the CAN bus was developed, there were no computer hackers,” Kitchen said. “If you study this data, it’s very easy to reverse engineer what the exact commands are to do various things in a vehicle, like put on the brakes, or disable the brakes,” he added.

Even if self-driving vehicles aren’t right around the corner, Silicon Valley’s tech talent and expertise in machine learning, coupled with advancements in laser and radar sensors, are driving the innovations needed to make autonomous vehicles a reality.

Stanford University engineering professor J. Christian Gerdes, who has been working on autonomous vehicle technology since 1992, concedes that significant challenges remain, from programming ethical decision-making to reduce crashes, to improving the performance of sensors in heavy rain or snow.

Still, he thinks the automated driving revolution is well underway.

“This is the most exciting time I’ve ever seen with automated vehicles,” he said.  “It’s no longer a question of can we make these cars, but a question of when will they impact our daily lives.”

This video story is featured in the QUEST television documentary, “Self-Driving Cars: The Road Ahead.”