More than a dozen RFQs for lidars are flying around. But automotive industry observers caution us not to expect a single lidar supplier to win this sweepstakes.
More than a dozen RFQs for lidars are reportedly flying around. Evidently, lidars are beginning to penetrate the ADAS market. Automotive industry observers, however, caution not to expect a single lidar supplier to win this sweepstakes.
Of all the automotive sensor modalities designed to enhance sensing and object detection, lidars offer by far the most complex and diverse options. Technologies deployed in each lidar system often differ not only in light sources, but also in their “ranging” and “imaging” methodologies. Some mature lidars are reaching mass production. But new lidar technologies — often pitched as more promising — are regularly appearing in the R&D or proof-of-concept stage. Inevitably, these lidars vary hugely in cost, size, and performance.
A further complication is the growing number of lidar players (estimated to be 70-80 companies) crowding the market. Mergers and acquisitions are rampant. Some lidar companies have found a new lease on life by opting for SPAC (special purpose acquisition company) status. Many startups view SPAC as an alternative way to raise capital and access public equity markets.
But the story gets even more complex. As lidar companies pursuing different technologies, automakers and Tier Ones are diverging in their vision of how to deploy a lidar for what applications.
“When we recently interviewed lidar companies, they told us that every OEM and Tier One has different requirements — demanding a specific field of view, distance and position to integrate lidars in a vehicle,” Pierrick Boulay, solid state lighting and lighting systems analyst at Yole Développement, told EE Times.
“This won’t be a winner-takes-all market,” Stephen Lambright, CMO of AEye, told us. Lambright noted that many lidar companies are still experimenting with numerous ideas, “much like a biotech industry.”
Of course, equating the lidar segment to biotech is hardly comforting.
But the automotive industry’s appetite for lidars is real. Lambright observed that in the ADAS world alone, his company is seeing 15 to 16 different lidar requests for quotation, all aimed at ADAS vehicles scheduled for production in 2025-26. Despite the diverging lidar market, Lambright noted, “Automotive companies are starting to place bets in the next 18 months on what they are deploying four or five years from now.”
In this update, EE Times reviews the current and future market landscape for lidars, covering existing design wins and a few new companies, such as AEye and Opsys, aspiring to become key players.
The best way to examine the technological variety in lidars is to draw a chart with two axes: one focused on imaging and another on ranging. Technologies used in imaging vary widely from mechanical, MEMS and optical-phased array to Flash. Ranging technologies include Pulse, FMCW (frequency-modulated continuous-wave) and phase shift.
Yole Développement mapped the lidar landscape in the accompanying chart by putting different players in different bins. The chart might not include everyone, as new players keep emerging, but this bird’s-eye view illustrates the diversity of technologies used in lidars.
Different light sources are used in various lidars. The majority of lidar manufacturers use edge emitter and avalanche photodiode (APD) at 905nm because those components are readily available in large volume. Some lidars, such as Luminar, have begun using a 1550nm all-fiber laser source. The challenge for the latter is cost reduction. However, the 1550nm laser is better at achieving longer range without compromising eye safety. The issue dogging 905nm lasers is that they are not eye-safe beyond low-resolution or short-range applications.
But edge emitters and fiber lasers aren’t the only options. Lidar companies are rapidly adopting vertical-cavity surface-emitting lasers (VCSELs), an idea sparked by Apple originally with iPad Pro 11’s lidar scanner. Ibeo in Germany, which uses VCSEL technology by Austria’s Ams, has a design win with China’s Great Wall Motors, scheduled for release in 2022. Ouster, which started out with mechanical lidars, announced last year a lidar for ADAS combining VCSELs and single-photon avalanche diodes (SPADs), with production scheduled in 2024. Opsys based in Israel is also developing a solid-state scanning lidar that uses a fully addressable VCSEL transceiver and SPAD receiver, slated for launch later this year. Opsys is promising a 200-meter detection range and scan rate at 1000 Hz.
Pros and cons
The course of lidar evolution runs from mechanical to MEMS, flash-based lidars and FWFC. However, this isn’t a simple straight line.
For example, many mechanical lidars still dominate today largely because they’re cheaper. The big drawback, however, is that mechanical components can become a source of failure, noted Boulay. The key advantage of MEMS lidar is its compact size. Its solid state makes it more reliable, said Boulay, but it still uses tiny moving parts.
In contrast, flash lidar comes with absolutely no moving parts. It is deemed more reliable, but the current generation still suffers from limited detection range, Boulay added.
Yole put FMCW in the R&D bin, calling it “TBD” — to be determined. Boulay said “We do not expect to see FMCW lidars before 2025.” Last year, Waymo talked about FMCW for its future home-grown lidars. Mobileye, most prominently, discussed FMCW early this year, describing it as its choice for lidars under development for fully autonomous vehicles.
Coherent detection is much more sensitive than direct detection, and it offers better performance, such as single-pulse velocity measurement and immunity to interference from solar glare and other light sources—including lidars used by other cars. However, serious challenges facing FMCW lidars include “manufacturing capabilities for silicon photonics” and the cost of technology, said Boulay. Mobileye, however, thinks it can overcome this by leveraging Intel’s silicon photonics expertise which includes its own fab, manufacturing, and IPs.
Design wins today
The best way to predict trends in the lidar market roadmap is to examine today’s design wins. Those announced include: Valeo-Audi’s A8; Valeo-Mercedes-Benz; Innoviz -BMW; Luminar – Volvo. While many wins mentioned here focus on installing one lidar in ADAS or fully autonomous vehicles, the newest wrinkle among car makers is to combine short, mid-range and long-range lidar in the same vehicle.
Good examples are: 4 lidars (1x Denso, 3x Continental) inside Lexus, two lidars (Velodyne, now transitioning to Valeo) in Hyundai; Valeo (x5 lidars) in Honda; and Ibeo (X3 lidars) in Great Wall Motors. Mix and match is the new trend. Lexus, said Boulay. It will use one long-range lidar from Denso, but add three medium-range lidars from Continental.
New lidar technologies getting adopted by ADAS vehicles are MEMS, fiber laser and SPAD array for flash lidars.
Here comes ‘Highway Autonomy’
Until several years ago, lidars were deemed a costly technology only affordable for a robo-taxi business. That’s no longer the case. Despite Elon Musk’s famous remark calling lidar “a fool’s errand,” lidars today are squarely on the map for new ADAS vehicles.
The ADAS world is seeing a rising demand for “highway autonomy,” judging by the RFQs from OEMs. AEye’s Lambright explained that OEMs want ADAS vehicles to allow drivers to take their hands off the wheel “with confidence.” He said, “We heard from OEMs that if you’re driving very fast on a freeway, such vehicles want to be able to detect small objects, tires, bricks, whatever it may be in the road, 150 to 200 meters away,” he explained.
While ranges less than 300 meters are good for many purposes, the ability to see farther with greater precision is desired for lidar application to improve performance and safety.
In highway autonomy that could fall into SAE’s Level 3 vehicle category where, Lambright said, the handoff between car and human driver “becomes critical.” He noted, “The farther you can see something out, the more time you can give the driver for transition and a safe solution.” Lambright described “small object detection at long ranges” as “the primary gauntlet that OEMs have laid down for us.”
In recent testing AEye placed Chevy Bolt and a Mercedes-Benz Sprinter delivery van at the other end of a runway. Watch for in the video, at 1018meters, Aeye’s lidar detected the Bolt and Sprinter van within a full field of view at 10 Hz.
VSI Labs, an applied research firm focused on the technologies used for active safety and autonomous control, recently oversaw AEye’s lidar performance testing in Byron California. VSI Labs’ validation report confirmed that AEye’s iDAR sensor could read dozens of points on an unmodified Chevy Bolt at 1,000 meters.
Phil Magney, head of VSI Labs, told EE Times, “We found it incredible to see a lidar detecting an object at a kilometer away. We’ve never seen anything like that.”
Comparing lidars against other sensors for highway autonomy, Magney said the problem is that “cameras don’t really see that far. Anything that comes from a camera is an estimate.” What about radar? “Well, radar is great but is really lousy at lateral positioning.” Lidar, however, “is an instrument that gives you a precise value on” what it sees.
But if an ADAS vehicle used for highway autonomy must change lanes, Yole’s Boulay made clear, “you need mid- to short-range lidars on the side.”
Given the variety of use cases envisioned for lidars, flexibility in architecture, wavelength and field of view becomes increasingly important.
Although Opsys isn’t the first company pursuing a high-performance SPAD and VCSEL lidar approach, its Micro-flash lidar, the company claims, can provide better than four times the range of flash LiDAR with superior resolution and scanning rate. In an interview with EE Times, Eitan Gertel, Opsys’ executive chairman of the board, promised a pure solid-state micro-flash lidar offering 200-meter range, capable of scanning full field of view at 1,000 frames per second.
“Opsys is younger than Ouster,” noted Yole’s Boulay, but with its solid background (the Opsys leadership team comes from Finisar, a manufacturer of optical communication components and subsystems), “they are experts in electric optical systems, and they know what they’re talking about.” Gertel was Finisar’s CEO between 2008 and 2015.
The key to Opsys’ micro-flash lidar is the bundling of multiple base sensors into a single lidar system, providing an integrated single 4D point cloud with a flexible FoV. This makes the micro-flash lidar “customizable,” said Gertel, for carmakers and Tier Ones seeking different field of views in different types of vehicles.
Even better, Opsys’ patented multi-wavelength technology allows multiple sensors to be installed in a vehicle and operate without interference.
Where to place a lidar?
Last but not the least, lidar location is a big issue for both cosmetic and practical reasons. Speaking of the company-owned autonomous vehicles it’s been testing on the road, “We use a conventional lidar installed on top of our vehicles,” said VSI Lab’s Magney. “As we’re driving around and it’s snowing or sleeting a little bit, that lidar is the first sensor to go.”
He explained, “A lot of it has to do with how much heat the sensor can build to melt the frozen matter. But when your lidar is on the roof, it is highly exposed to weather. Then, sensor cleaning becomes an issue.”
When evaluating AEye’s lidar performance, VSI Labs was impressed that AEye can put the lidar behind the windshield, which has very little impact on the lidar performance. The resulting impact is in less than a 10 percent degradation in performance, said AEye. Given that glass, especially a windshield, is very dispersive, AEye’s Lambright suspects that it could create “a tremendous amount of interference” for frequency oriented lidars such as FMCW.