Quanergy Announces $250 Solid-State LIDAR for Cars, Robots, and More
Velodyne Announces a Solid-State Lidar
Lumotive Says It’s Got a Solid-State Lidar That Really Works
In 2015 Quanergy wowed crowds at CES with claims that it had a lidar that steered its laser beam without moving parts. Such a solid-state lidar would be smaller, tougher, and cheaper than the handmade rotating roof towers of Velodyne, then as now the leading lidar company.
A year later, when Quanergy cited a price of US $250, the top-of-the-line Velodyne lidar was selling for around $75,000. Investors valued Quanergy at $1 billion, and Mercedes announced a partnership.
But manufacturing problems set in. The company began walking back its claims for range and for price, outside testers faulted the lidar’s accuracy in real-world applications, Mercedes distanced itself, and key personnel jumped ship. It was a classic case of hype and its comeuppance.
But the idea of a true solid-state lidar retains its appeal. It would stand up to automotive conditions, laughing off potholes and sneering at harmonic vibrations. It would be small enough to disappear into the grillwork. And it could be alluringly cheap if only the manufacturing volume were to rise high enough.
Now Louay Eldada, Quanergy’s chief executive, tells IEEE Spectrum that his company is poised to reclaim its leading position in lidar. Reason: Its fabrication troubles are pretty much behind it.
“Our yields today are over 97 percent, we’re working on getting them well over 99 percent, and then we’ll be adding more 9s,” he says, attributing the improvement to better machine control. “We pick each chip, do a quick measurement, and place each chip in the chip stack in automated fashion. Pick, measure, place—it takes 2 seconds.”
It’s the first time he’s released those yield numbers, he says. But his specs for the lidar’s range—150 meters for the farthest-looking model—are the same he cited to Bloomberg News last August. And as Bloomberg noted, that’s below the industry standard of 200 meters for driving at highway speeds.
The price for the longest-range lidar in its lineup is “under $1,000,” Eldada says, and as little as $500 if manufacturing volume rises enough. That’s not bad, but it’s a far cry from $250.
Eldada touts his device’s minimum range: zero. That’s what you need to see a child dart out from behind a bush to retrieve a lost ball. A device using MEMS mirrors can’t see a child at such close range because it uses the same oscillating mirror to send light out and collect the returning echo, and an echo off a close object will overwhelm the system. Quanergy, however, uses separate components to send and receive the laser beam, and they’re isolated well enough to avoid crosstalk.
Eldada worked on the core technology for his Ph.D. dissertation at Columbia. It does for light what phased array radars did for microwaves, forming a new beam out of the interference pattern of progenitor beams.
It works like this: You break a beam into daughter beams, then delay one of the daughters by passing it through a medium that’s under the influence of some moderator, like an audio or an electrical pulse (Eldada won’t say what moderation he uses). That delay shifts the phase, so that one beam’s peaks and troughs come after those of the other beam, interfering. The interference then constitutes a new beam, one that can be steered by playing with the phase delay.
Doing all that on a stack of chips requires extremely precise alignment. And that was the source of the manufacturing problem that Eldada claims to have solved, or nearly solved.
Lumotive, a startup that recently came out of stealth mode, claims that it, too, uses a true solid-state technology. But it manages the phase interference differently, without on-chip waveguides that must be fit together precisely. It also claims that its fatter beam can carry more energy while remaining eye-safe, and that makes for greater efficiency and range. Oh, and Lumotive says that it, too, can see what’s in front of its nose, centimeters away.
Of course, there’s no way for an outsider to check these claims, or to compare the two companies’ different ways of managing an optical phase array. It does bear noting, though, that neither lidar company has been publicly backed by a car company.
Today some Quanergy lidars are in use for security purposes. Some are being sold to Orion Entrance Controls to help screen people who enter and leave buildings. Another application might be in the proposed wall between Mexico and the United States, a project in which Quanergy has expressed interest.
There’s no particular security advantage in being solid state, Eldada acknowledges. But it’s cheaper than the alternative of mechanical lidars.
Meanwhile, the long-sought automotive application, in partially-automated cars that hew to “Level 3” autonomy, lies just beyond the next fork in the road. As it always seems to do.
“We’ll start to roll out in 2021,” Eldada says. “We will make some sort of announcement this year.”