Google's Self-Driving Car Software

On paper, Google’s plan for self-driving cars reads something like this: Step One: Build cars that can navigate flawlessly without a human at the wheel. Step Two: ????? Step Three: Profit According to Google, Step One is proceeding rather well. “A mile of city driving is much more complex than a mile of freeway driving, with hundreds of different objects moving according to different rules of the road in a small area,” read an April 28 posting on Google’s official blog. “We’ve improved our software so it can detect hundreds of distinct objects simultaneously—pedestrians, buses, a stop sign held up by a crossing guard, or a cyclist making gestures that indicate a possible turn.” The Google cars’ onboard computers can reportedly deal with everything from train crossings to cyclists to construction work, and react accordingly. Google has used data from 700,000 miles’ worth of driving to build software models that can, for example, recognize that a cyclist with an arm extended is about to change lanes. “We still have lots of problems to solve, including teaching the car to drive more streets in Mountain View before we tackle another town,” the posting added, “but thousands of situations on city streets that would have stumped us two years ago can now be navigated autonomously.” If Google can perfect its self-driving cars, the result could disrupt a number of automotive-centric industries. Just as robotics and automation had a seismic impact on manufacturing, trucks and cars capable of self-navigation may alter how companies deliver packages and ferry people around. That won’t necessarily turn out great for everybody—picture human truck- and taxi-drivers protesting the new fleets of self-driven vehicles—but it could create a whole new industry for software programmers and developers with experience in automotive systems. Nor is Google the only company exploring the possibilities of self-driving cars. Late last year, Tesla Motors posted a job for an autonomous driving engineer; tasks included designing algorithm validation methods to reduce development time. Nissan, Audi, and Mercedes are also exploring ways of taking the driver out of the equation, or at least increasing the role of computers in navigating a car from Point A to B. Now Google just needs to figure out Step Two.

Related Stories

Image: Google