Austin Russell, Luminar’s CEO and co-founder, is convinced his company has the most advanced, highest resolution laser LIDAR sensor on the market and he’s quickly boosting production to supply autonomous vehicle test fleets.
One year after emerging from stealth mode Luminar, a maker of laser LiDAR sensors for self-driving vehicles led by 23-year-old optics whiz Austin Russell, says it’s ready to build an upgraded version of its high-tech sensor in high volume.
Luminar’s latest device is lighter, smaller and needs less electricity to generate 3D “point cloud” maps of cars, pedestrians, bicyclists, dogs, trees and other objects robot cars encounter, Russell said. Though it uses just a single laser (industry leader Velodyne’s latest unit has 128 beams), Luminar boasts image quality that’s as good or better than any rival’s LiDAR, and design tweaks now allow its sensor to be built much faster.
“We’ve gotten the build time down from what effectively took a full day of an optics Ph.D.’s time to assemble. … Now we can build each unit with build-time of about eight minutes each,” Russell told Forbes. “By the end of this year, we’ll be at a production rate that’s north of 5,000 units a quarter.”
Another key change: A huge cost reduction for the exotic indium gallium arsenide receiver chip Luminar uses to process laser signals. Acquiring chip designer Black Forest Engineering in Colorado Springs helped cut the cost of that chip from “tens of thousands of dollars each to about $3,” Russell said, without sharing current pricing details.
Those advances and being able to supply LiDAR sensors in high volume is vital for the young company as robot car fleets grow exponentially, from a few thousand worldwide currently to many times that number in the next few years. Velodyne has dominated the market for years, but competition has mushroomed recently with scores of new players including Quanergy, LeddarTech, Ouster, Innoviz, TetraVue, AEye, a host of Chinese startups and numerous others claiming performance or cost advantages rushing to lock in lucrative supply deals with auto and tech firms.
Velodyne’s big advantage, aside from a decade of making LiDARs that range from a few thousand dollars to more than $70,000 each, is its ability to produce vastly greater quantities than any rival. And it’s not slowing down. CEO and founder David Hall, who like Russell also invented his LiDAR sensor, is working to raise Velodyne’s annual output to a million units a year.
Waymo, the Alphabet Inc. unit that is launching a commercial self-driving ride service in the Phoenix area this year, makes its own LiDAR sensors. The company this year settled a bitter legal fight with Uber, over allegations that a former Uber executive stole trade secrets related to Waymo’s proprietary LiDAR designs.
Closely held Luminar raised $36 million in its initial funding round last year, and now has 350 employees split between its Palo Alto, California, headquarters and Orlando, Florida, factory. Currently, it has four automotive partners, though it’s only identified Toyota Research Institute as one. The other three are “of similar stature,” Russell said, without elaborating.
Luminar’s production volume will be sufficient to supply every major AV program, he said. “There are a ton of programs that are really excited to deploy our platform at scale.”
An acronym for Light, Detection and Ranging, LiDAR’s ghostly images are essential to the advanced sensor rigs used by self-driving vehicles. In daylight, cameras are great at detecting objects and colors, while radar can sense dense objects at great distances. But combining them with LiDAR’s ability to see virtually all objects in three dimensions, in daylight or at night, gives robotic vehicles computer vision that’s more powerful than what the eyes of human drivers can see. And with the notable exception of Tesla’s Elon Musk, every major autonomous vehicle program in the world relies on LiDAR.
That’s what makes the fatal accident in March in Tempe, Arizona, in which an autonomous test vehicle operated by Uber struck and killed a pedestrian who was slowly walking her bicycle across a dark city street, so troubling. In the first of its kind accident, Uber’s Volvo XC90 had both LiDAR and radar sensors and yet the vehicle didn’t stop for the pedestrian in time.
It’s not yet known whether the sensors failed to function correctly, if the LiDAR unit was positioned in a suboptimal manner on the Uber vehicle or whether Uber should have been using multiple LiDARs on the vehicle, something most other AV test fleets do. Federal investigators dispatched to determine the cause haven’t yet released their findings.
Luminar gets its high range and resolution by using a higher-cost 1550-nanometer laser, versus the relatively cheap 905-nanometer beams used in more conventional designs, said Alex Lidow, CEO of EPC, a supplier of electronic components used by LiDAR makers. Likewise, it’s receiver chip has been much pricier than those used in Velodyne models, though the cost reduction made there now brings it into line with key competitors, he said.
“Luminar is doing something a little bit different, and in the end we’ll find out if they’ve got it right.”
LiDAR maker Luminar says its sensors generate the highest-resolution images currently available for autonomous vehicles.