Google Spinoff Waymo Has Built Its Own Self-Driving Sensors

DETROIT (The Washington Post) -
Waymo CEO John Krafcik unveils a Chrysler Pacifica Minivan equipped with a self-driving system developed by the Alphabet Inc. unit at the North American International Auto Show in Detroit, Michigan, January 8. (Joe White/Reuters)

Google’s autonomous driving spinoff, Waymo, has developed sensors that pair with its self-driving software, potentially opening the door for the company to sell a comprehensive system that automakers build into future car models.

Google initially built its self-driving software on a prototype car outfitted with sensors, cameras and other hardware from outside suppliers. But to build a more affordable and sophisticated system capable of fully autonomous driving, the company decided it needed to create both halves of the technology, executives said.

The announcement comes just weeks after Japanese automaker Honda said it would incorporate Waymo’s technology into some of its vehicles. The companies said that deal was centered on research rather than producing vehicles for market, Bloomberg News reported.

Chrysler has already adapted its Hybrid Pacifica minivan to use Waymo’s self-driving hardware and software, and one of those vehicles was put on display for the news media Sunday. This month, a fleet of 100 will begin driving on public roads in California and Arizona.

Waymo chief executive John Krafcik did not outline specific plans for selling the self-driving sensors in remarks prepared for a keynote address Sunday at AutoMobili-D, an automobile technology conference that began the day before the North American International Auto Show is set to start here.

“Today we can imagine our self-driving technology being applied in a variety of ways: personal transportation, ride hailing, logistics, and public transport solutions just to name a few,” Krafcik said in prepared remarks.

Krafcik also reported Sunday that engineers had to take control of the car while driving in autonomous mode less often in 2016 than the previous year, an indicator that the technology has improved. In 2015, engineers intervened 0.8 times for every 1,000 miles. In 2016, that number fell to 0.2 times.

It has been previously reported that Google was rolling back plans to build its own car, although the company initially developed a self-driving prototype, called Firefly, that had no pedals or steering wheel. The company has touted the potential for self-driving technology to reduce fatal car collisions and assist those incapable of driving on their own.

“To solve these challenges, we’re thinking bigger than a single use case, a particular vehicle, or a single business model,” Krafcik said in his prepared statement.

Google spun its self-driving car project into a stand-alone company called Waymo last month, a sign that parent company Alphabet is serious about bringing the technology to market. The company also revealed that it quietly completed the first-ever ride with no driver in the vehicle in October 2015, when a blind passenger took a trip around Austin.

Google set out to create self-driving technology in 2009, and Krafcik said the sensors unveiled Sunday are the latest iteration of that research and development.

The system Waymo has developed surveys the car’s surroundings in three ways. The car can “see” its 360-degree surroundings and navigate accordingly using LIDAR sensors that send out millions of laser points each second and eight “vision modules” that construct a high-resolution picture. A radar system tracks objects as they move around the car.

“It’s a virtuous cycle,” Krafcik said in prepared remarks. “Better hardware gives us better data to develop our software; and as our software become more sophisticated, we get better at optimizing the most important aspects of hardware.”

The company has tested its technology on nearly 2.5 million miles of road to date.