Driverless Cars

Robots Are Taking the Wheel

By | Updated July 8, 2016 3:58 PM UTC

The auto industry seems to be hurtling headlong into an era when cars drive themselves. It may sound futuristic, but parts of it are already here. The robots are easing us out of the driver’s seat bit by bit. Cars on the road today can brake for you or steer you back into your lane, while others coming in the next two years will change lanes automatically and allow you to drive hands-free. That makes plenty of people nervous, and there’s no end to the ethical and legal questions yet to be settled. But there’s little doubt that the robot driver’s day will come. For one thing, carmakers see a future in which autonomous “transport pods,” often provided by ridesharing services like Uber or Lyft, supplant the tradition of two owner-operated cars in every garage. For another, we’re no longer swarming toward suburbia, the defining demographic trend of the 20th century. Instead, we are heading toward cities by the millions. So we have a choice between mega-gridlocked mega-metropolises, and driverless cars that move in harmony like schools of fish.

The Situation

The leading automakers are pouring billions into developing autonomous vehicles that use sensors, cameras and high-speed computing power to read and react to traffic, pedestrians, stoplights and infrastructure. Luxury lines lead: BMW is rolling out a car that can park itself, Cadillac has a model coming that drives hands-free on the highway, while Mercedes and Audi already offer models that can pilot through a traffic jam while only asking its human minder to touch the steering wheel occasionally. Tech giants Google and Apple are also in the race. Google and Fiat Chrysler Automobiles are teaming up to develop about 100 self-driving Chrysler minivans, while BMW is collaborating with Intel and Mobileye, an Israeli maker of components for autonomous systems. GM is investing in Lyft and is developing a fleet of robot taxis. GM also invested $1 billion in self-driving startup Cruise Automation and Ford put $182 million into cloud computing startup Pivotal Software, which helped it develop a mobility app Boston Consulting Group estimates that robot cars may account for a quarter of global auto sales by 2035. Fully autonomous vehicles may be navigating cities in five to 10 years.

 

The Background

The dream of a self-driving car first appeared in the pages of science fiction and then in the General Motors Futurama display at the 1939 New York World’s Fair. Computing power didn’t catch up with our imaginations until the 1980s, when Carnegie Mellon University came up with a robot Chevy van and Bundeswehr University Munich developed an autonomous Mercedes van. Consumers got their first taste of autopilot in the 1990s when Toyota, Mitsubishi and Mercedes began offering adaptive cruise control, which uses radar to automatically adjust vehicle speed to keep a set distance from cars ahead. As the cost and size of the sensors and chips have plunged, autonomous features have proliferated and can now be found in everyday Hondas and Fords. Google accelerated the pace of development by logging more than 2 million miles testing its driverless cars on Silicon Valley roads. In initial road tests, driverless cars actually have had an accident rate twice as high as human-driven models — though researchers says it’s the humans in other cars who are generally to blame.

The Argument

The accelerated pace of invention is outstripping government’s ability to regulate self-driving cars. There are few laws on the books requiring drivers to keep their hands on the wheel because no one imagined that would be an issue. The first cars capable of talking to each other to warn of traffic hazards and keep a safe distance will hit the road in 2016. Yet safety regulators are still working on rules for how vehicle-to-vehicle communication should work. The U.S. plans to expedite new regulations and has proposed spending $4 billion over 10 years on research and infrastructure. Federal safety regulators have ruled that Google's self-driving software is the equivalent of a human driver. More broadly, there’s the ethical dilemma of turning over decision-making power to a robot in a life-or-death situation. The question of liability also remains unanswered. When a car on auto-pilot causes an accident, who is at fault? Automakers also have yet to design a connected car that cannot be hacked, raising security concerns and dystopian scenarios of robot cars run amok. Driverless cars could save thousands of lives in developing nations, where accident rates are high, and be a boon for the disabled or the elderly. But a fatal accident involving a Tesla sedan driving on autopilot was a reminder of the difficulties in getting to a hands-off future. 

The Reference Shelf

 

 

First published June 24, 2015

To contact the writer of this QuickTake:
Keith Naughton in Southfield at knaughton3@bloomberg.net

To contact the editor responsible for this QuickTake:
John O'Neil at joneil18@bloomberg.net