The 2020 Consumer Electronics Show was absolutely crawling with companies hawking lidar. Short for light radar (yes, really), this powerful type of sensor generates a three-dimensional pointcloud of its surroundings. Experts and industry insiders not named Elon Musk see it as a key technology for self-driving cars. There are dozens of companies developing lidar technology, and each insists that its sensor is a cut above the rest.
But while every lidar is above-average in the halls of CES, things are starting to look different in the real world. At least one segment of the market—custom robots for warehouses, mines, and other industrial sites—is starting to buy lidar sensors in significant volume. Another segment—low-end lidars used in car driver-assistance systems—is poised to become a big market in the next couple of years.
For this piece I asked both lidar company officials and independent experts to help me understand the state of the lidar market. They told me that Velodyne—the company that invented modern three-dimensional lidar more than a decade ago—continues to dominate the industry.
But Velodyne is facing growing competition from newer firms. One company in particular—Ouster—has begun shipping aggressively priced alternatives to Velodyne’s flagship products. While these products might not quite match Velodyne’s performance, they’re good enough and cheap enough to pose a serious threat to Velodyne’s dominance.
The big battles in the lidar market are still in the future. A lot of lidar sales so far have come outside of the automotive industry, but experts expect carmakers to be the biggest customers for lidar. In the next few years, we’re going to see a number of carmakers make their first bulk lidar purchases—buying thousands of low-cost lidar sensors to improve their advanced driver assistance systems (ADAS). A number of lidar companies are positioning themselves to win these deals, and some are pairing up with traditional “tier 1” automotive suppliers to improve their odds.
The industry’s biggest prize may be supplying more powerful lidar sensors for use in fully self-driving vehicles. Many companies are angling to serve this market, but those sales are still quite a way off because fully self-driving technology isn’t yet ready for prime time.
Spinning mechanical lidar still dominates in robotics
Velodyne invented the modern three-dimensional lidar industry more than a decade ago, and it continues to be an industry leader. Its original lidar had 64 lasers, stacked vertically, that were spun around to give a 360-degree view of the surroundings. Velodyne has since introduced variations on this basic design, including a high-end 128-laser design and cheaper models with 32 or 16 lasers.
Velodyne’s top-of-the-line lidar units traditionally cost $75,000 or more. Until recently, people who needed high-performance lidar had little choice but to pay up. But in the last couple of years, a startup called Ouster has started to give Velodyne some much-needed competition.
The classic Velodyne lidar had 64 individually packaged lasers inside, each paired with an individual light sensor to detect return flashes. This complex design contributed to the units’ high price. By contrast, Ouster uses semiconductor-based technology to pack 64 lasers onto a single chip, with 64 detectors packed on a second chip. This reduces the cost of Ouster’s lidar in much the same way that the microchip revolution enabled the creation of cheap personal computers in the 1970s.
The result: in 2018 Ouster was able to offer its first lidar sensor, a 64-laser unit called the OS-1, for just $12,000. That was dramatically cheaper than Velodyne was charging for its 64-laser models at the time. Since then, Ouster has expanded its product line. The company’s products now range from a low-end 16-laser unit for $3,500 to a long-range unit with 128 lasers for $24,000.
“The fact that you can get a 64-channel spinning lidar for $12,000 was unheard of,” said John Williams. “When we realized it existed, we were surprised.”
Williams is the chief technology officer at Kudan, a company that makes software to help robots track their own location (a problem known as SLAM in the robotics world). Kudan customers build robotic products like low-speed shuttle buses, warehouse supply carts, autonomous forklifts, and so forth. Kudan’s software needs to work with a variety of sensors, including cameras and lidar, so Williams has good visibility into which lidar sensors robotics companies are using in the real world.
Williams told Ars that most of his clients are still buying Velodyne lidars. But “Ouster is the up-and-comer in terms of disrupting the market,” Williams said. “In terms of accuracy and noise,” Ouster sensors are “not quite as good as Velodyne,” Williams said. But they’re dramatically cheaper, and that matters to many Kudan clients. Clients who wouldn’t have considered buying lidar at Velodyne prices reconsidered once Ouster introduced a 64-laser unit for $12,000.
When I asked Velodyne executive Rick Tewell about Velodyne’s prices, he insisted that they were “extremely competitive.” But he wouldn’t give me specific numbers, and he didn’t seriously dispute that Ouster’s high-end lidars were cheaper than comparable offerings from Velodyne.
Velodyne also faces competition from some Chinese companies that Velodyne dismisses as copycat vendors. Like Ouster, these companies offer Velodyne-like sensors at low prices. But Velodyne says that unlike Ouster, these companies simply copied Velodyne’s design. Indeed, Velodyne sued two of those vendors—RoboSense and Hesai—for patent infringement last August.
In recent years, there has been no shortage of hype about the potential of solid-state lidar—lidar that’s fixed in place rather than spinning 360 degrees. There are dozens of companies working on lidar sensors, and most of them are solid-state designs. Yet they don’t seem to be getting much use in the real world—at least not in the commercial robotics market. Williams told Ars that “we haven’t come across solid state in the wild yet.”
That’s partly because a number of solid-state companies haven’t actually started shipping products to the general public yet. It’s also partly because some lidar companies are focused on large-volume sales to carmakers, not one-off sales to smaller robotics companies.
But it’s also because spinning lidar has some unique advantages. Most obviously, a single spinning lidar sensor offers 360-degree coverage around a vehicle. To achieve similar coverage with a fixed lidar system you need multiple sensors distributed around the vehicle. That means higher costs and power consumption.
Some lidar companies are fighting for the ADAS market
Right now, most of the companies actually shipping lidar-based products are outside the automotive industry. But experts expect the car industry to become a hugely important market for lidar companies in the next few years.
The automotive lidar market can be broken down into two segments. At the high end, there are companies working on fully self-driving technology—dubbed “Level 4” autonomy in industry jargon. These technologies are likely several years away from commercial deployment, so companies building fully autonomous vehicles are only buying a few lidars at a time for research purposes.
These tend to be well-funded projects that are just trying to get their self-driving stacks to work. Cost isn’t a major concern for these companies; they’re perfectly willing to spend tens of thousands of dollars if that’s what it takes to achieve full self-driving capabilities. Many of these companies are also planning to build autonomous taxis, not customer-owned cars. This leaves more room to buy high-priced hardware because the costs can be spread out over thousands of customers.
At the opposite end of the market are car companies planning to use lidar to enhance existing advanced driver-assistance systems (ADAS). These companies are highly price-sensitive, since they have to pass the cost of a lidar on to customers. Experts believe lidar for the ADAS market needs to cost less than $500 for the economics to work. But if lidar companies can hit the right price level, the potential market is huge. By the middle of the decade, carmakers could be selling millions of lidar-equipped cars.
Until recently, many people assumed that fully self-driving cars were just a few years away, and hence there would be a large market for lidar with long range and high resolution.
But the expected full-autonomy breakthroughs have been slow in coming. Companies that had aimed to launch self-driving taxi services in 2019, 2020, or 2021 have been forced to revise their timelines. And that has created a headache for lidar companies that were aiming to supply lidar sensors for fully autonomous vehicles.
“A lot of these companies are trying to shift their focus right now,” said Sam Abuelsamid, an automotive industry analyst at Navigant. “There’s a lot more emphasis on trying to get lower-cost sensors out there and get them into ADAS applications,” he told Ars.
Beyond the slow progress of fully autonomous vehicles, Abuelsamid argues that European regulators are also driving the adoption of lidar. A government-backed rating agency called Euro NCAP is expected to require cars to have more sophisticated advanced emergency braking systems to earn a five-star safety rating starting around 2023. Adding lidar could help carmakers clear the bar.
The companies best positioned to sell lidar to carmakers are likely to be those carmakers’ established suppliers. Bosch, one of the world’s biggest “tier 1” automotive suppliers, recently announced its own lidar sensor, though it’s light on details.
Another significant player may be Ibeo. A few years ago, Ibeo sold the car industry’s first production lidar to Audi in partnership with tier 1 supplier Valeo. More recently, Ibeo sold a 40-percent minority stake to a different tier 1, ZF Friedrichshafen. Ibeo and ZF are now developing a new generation of lidar with more than 10,000 lasers on a chip. Like Ouster, Ibeo is planning to create arrays of lasers using conventional semiconductor chip techniques. They believe this will enable them to sell the devices for a few hundred dollars. They’re hoping to score deals with carmakers and be ready for volume production in late 2022 or early 2023.
Velodyne wants to get into this market, too. At the 2020 CES show, Velodyne announced a new sensor, called the Velabit, that the company says will cost around $100 in volume. If Velodyne can hit that price and deliver reasonable performance, it could be a compelling option for cost-conscious automakers.
The fully autonomous vehicle market remains wide open
We’ll start to see clear winners and losers in the low-end automotive market in the next couple of years, as automakers line up with preferred lidar providers. But it’ll take longer for things to shake out in the market for the more powerful sensors required for full autonomy.
A bunch of lidar makers have announced partnerships with major automakers or others working on fully self-driving car technology. But it’s difficult to tell how significant these deals are. If a lidar company manages to sell a handful of test sensors to a car company, that doesn’t necessarily mean the carmaker will buy thousands or millions of units from them later. The evaluation units lidar companies send to potential customers may not even be ready for mass production. In this market, the line between a research prototype and a shipping product is pretty fuzzy.
There are many, many lidar companies vying for this market. Besides Velodyne and Ouster, which I’ve already mentioned, these include Aeva, AEye, Baraja, Livox, Luminar, Sense, and a number of others. There are also several companies with sophisticated lidar technology that have been acquired outright by self-driving companies.
Ford’s Argo, GM’s Cruise, and self-driving startup Aurora have all acquired lidar companies in the last three years. Alphabet’s Waymo has long had a team developing its own lidars. The Russian search engine company Yandex has an ambitious self-driving project, including its own lidar team.
I detailed several of these companies’ technological approaches in a piece last year. But in a nutshell, the companies are experimenting along a number of different dimensions:
Beam steering. Every lidar sensor needs some way to point laser beams in different directions. A few companies are following Velodyne’s strategy of mounting lasers on a spinning gimbal. Others are using a tiny, mechanically adjustable mirror to steer a single laser beam in multiple directions. Still others, including Baraja, are passing laser light through a prism; this allows the laser’s direction to be controlled by its wavelength.
Distance measurement. Most lidar sensors send out a laser pulse and then time how long the light takes to bounce back. But a few, including Aeva and the recently acquired Blackmore, use an alternative approach called frequency-modulated continuous wave (FMCW) lidar. This approach continuously varies the frequency of the outgoing laser beam, which is split in two as it’s sent out into the world. When the light bounces back, the two beams are recombined. This creates a beat frequency that depends on how far the first beam traveled, allowing a precise distance measurement.
As an added bonus, FMCW lidar systems can use the Doppler effect to estimate an object’s speed as well as its distance. Baraja uses a variant of FMCW that randomizes the outgoing frequency. The company says this technique makes its lidar highly resistant to interference.
Laser wavelength: Most lidars transmit at wavelengths at or near 905 nanometers, just outside the visible range. A big advantage of these wavelengths is that they work well with conventional silicon-based technologies, making it fairly easy to create low-cost chips. The downside is that the fluid in the human eye is transparent to these wavelengths, creating the risk that a high-powered laser could burn the human retina. To avoid that, regulations strictly limit the power levels of lasers at or near 905nm.
Some companies, including Luminar, AEye, and Baraja, have opted to use 1550nm lasers instead. The fluid in the human eye absorbs light at 1550nm, minimizing danger to the retina. That gives these companies an option to pump considerably more power into their lasers, which makes it easier to achieve long ranges. This has a couple of downsides, however. One is that working with 1550nm lasers requires more exotic semiconductor materials, driving up costs. And higher power levels can be a downside in its own right, since self-driving cars have a limited power budget.
After talking to executives at most of these lidar companies and a number of independent experts, I don’t know which of these approaches will be successful. To win the high-end lidar market of the future, a company will need the right combination of range, accuracy, cost, power consumption, reliability, and other factors. Some of these lidar companies aren’t close to having shipping products, but more importantly, few if any of their potential customers in the self-driving sector are close to launching commercial products either.
But I continue to be confident that some of these companies will figure out how to build lidar that’s powerful, reliable, and cheap enough for use in mass-market self-driving cars. Companies aren’t just exploring ways to make their sensors more powerful and reliable—they’re also working on ways to bring down costs. Some are planning to leverage the economies of scale found in the consumer electronics industry. Others are planning to leverage the integrated photonics techniques pioneered in the telecommunications sector.
And all of these companies say that their products will get dramatically cheaper when they’re sold by the thousands or the millions instead of one unit at a time. It’ll take a few years to get there, and the winning company might be someone who isn’t even on my radar yet. But it’ll happen—and probably sooner than skeptics think.