Grand Theft Autonomous: Google Accuses Uber of Stealing LiDAR Design
Author: Alex Webb
Spacecraft use it to measure distance. Farmers use it to work out which fields need fertilizer. Archaeologists use it to map topography. And, crucially for Uber Technologies Inc. and Alphabet Inc.’s Waymo, self-driving cars use LiDAR to navigate.
As carmakers and technology entrants scramble to develop autonomous vehicles, LiDAR has become a highly coveted technology. And now it’s at the center of a lawsuit pitting Waymo against Uber, the ride hailing upstart seeking to create its own autonomous vehicle empire.
Photo: Waymo
Waymo, the self-driving car business of Google parent Alphabet, accused Uber in a court filing on Thursday of stealing its LiDAR designs. Former Google engineer Anthony Levandowski downloaded 9.7 gigabytes of files before he left to found a startup that was subsequently acquired by Uber, the company said in the suit.
“Misappropriating this technology is akin to stealing a secret recipe from a beverage company,” Waymo said in a blog post on Medium. “We believe these actions were part of a concerted plan to steal Waymo’s trade secrets and intellectual property.”
"We take the allegations made against Otto Motors and Uber employees seriously and we will review this matter carefully," Uber spokeswoman Chelsea Kohler wrote in an e-mail.
A look at the technology at work in LiDAR -- and the role it plays in helping cars drive by themselves -- helps explain why the companies consider it so vital.
Image: Waymo
LiDAR is a radar-like system that uses lasers instead of radio waves to build a 3-D image of the surrounding landscape. Since satellite navigation systems are only accurate to within 16 feet and can be easily flummoxed by high-rise and glass-fronted buildings, autonomous vehicles require an array of other sensors to position themselves precisely and maintain awareness of nearby pedestrians, vehicles and other objects.
Waymo has its own LiDAR. LiDAR specialists such as Velodyne LiDAR Inc. say that, to cut the driver completely and permanently out of the equation, the technology is essential. The Morgan Hill, California-based company counts Tesla, Uber, Google and carmakers such as Ford Motor Co. and Volkswagen AG among its customers. Those companies use LiDAR for research and mapping purposes. In the models that are on the road, Tesla Inc. doesn’t use a LiDAR, instead deploying a combination of image cameras, sonar and classic radar for its Autopilot system, which allows limited autonomous driving.
LiDAR comprises a series of rotating, stacked lasers that shoot out at different angles. Each layer is called a channel, and is made up of two laser beams. The signal from each individual channel creates one contour line, and together, those lines generate a 3-D image of the surrounding environment. That means that, the more lasers in each stack, the higher the resolution. Velodyne, for instance, manufactures products with 16, 32 and 64 laser channels.
The main hurdle to LiDAR becoming a widely adopted technology in mass-produced cars is cost. A 64-channel unit from Velodyne can cost more than $50,000, while the lower end 16-channel product sells for $7,999. Since a car might require several LiDAR units, it quickly makes the cost prohibitive for anything but the most expensive luxury cars. Velodyne and competitors such as Quanergy Systems Inc. are working to reduce the price. That would be accelerated by major orders for mass market cars.
The appeal of LiDAR has prompted a race for the technology. Automotive supplier Continental AG bought a LiDAR business from Advanced Scientific Concepts Inc. in 2016, while France’s Valeo SA has teamed up with Canada’s LeddarTech to supply the product.