Lidar Technology review

[Dietl]

This forum is dedicated to technology review in lidar. Most posts pen by myself will be internet gathered pieces of data. I will credit authors with names posted and link. Hopefully new members will add to this.

[Dietl]

Primer,

Taken from internet - unknown author

 

LiDAR - The Science

LiDAR (Light Detection and Ranging) is the process of bouncing a beam of light from an object to the source and measuring how long it takes to return. That means there are a few key components: a light (laser) source, a receiver, and some form of an optical unit that steers the beam

Micro-Electro-Mechanical Structures (MEMS) - In this approach, a laser is reflected off tiny, microscopic mirrors on a chip that steers the laser beam around.

Optical Phased Array - This is similar to how noise-canceling headphones work; a sound wave is canceled by emitting a second wave where the peaks of the first line up with the valleys of the second, so the waves add to zero. Light is an electromagnetic wave (the wave's frequency determines the color), so the idea here is to use a similar process. Instead, using an array of wave emitters (sources) that interfere or 'add' up so that they are canceled in some places but build on each other (peaks line up with peaks) in other places--in a beam for LiDARs. This technology is more complex and depends heavily on the laser source.

Digital LiDAR - This isn't necessarily a steering technology; the idea here is to have an array of laser sources and a diverging optical lens (so lasers are fired straight, but lasers further from the center of the lens are angled more). The lasers are then fired one at a time or a column, and the lens causes each laser beam to go in a different direction.

Flash LiDAR - it's essentially just a floodlight. Instead of trying to steer a light source around to illuminate a source, we'll illuminate everything at once (like a camera flash) and see what we can pick up.

Now for the light sources. There are many ways to do this, so I'm just going to focus on two general LiDAR approaches, continuous wave and time of flight, since the approach largely dictates the parameter of your source.

Time of Flight - The idea here is that a laser is pulsed, and distance is measured by the time between the pulse and the reflection of the pulse. The laser sources are more straightforward; you only need to emit at one frequency (or color when we're thinking about light).

Continuous Wave (or Frequency Modulated Continuous Wave) - A laser is always on instead of pulsing on and off in this approach. Still, its frequency (or, again, color when we're thinking about light) is changed. Distance is measured here by a frequency and phase shift. Essentially the receiver can tell the frequency emitted a signal is measured. The laser sources here are much more specialized; they need to change their frequency continuously.

Mechanical lidar

Limitations: cost and reliability. Cost is mainly due to the manufacturing process and many parts. The high-resolution 128-channel, meaning 128 individual lasers that spin around, is priced between $20000 and $75000. As of 2018, Velodyne isn't super open about prices). Reliability is because these things are constantly spinning, and slip rings wear out.

 

Benefits: Besides providing a 360-degree field of view, the most significant benefit to mechanical LiDARs that you've never thought of is eye safety and power. Regulations that limit the power a LiDAR can operate are essentially based on how long someone can stare at your sensor without significant eye damage. Since mechanical LiDARs spin, they're only facing a person for a short amount of time and so are allowed to operate at much higher powers (which increases maximum range)

MEMS

Limitations - vibration, reliability, and efficiency. The first is pretty straightforward; when you vibrate a set of tiny mirrors, things get thrown out of alignment, and measurements get dropped. It is also coupled to the reliability problem; in a lab, MEMs are pretty reliable, but their lifespan is shorter in real-world scenarios. Finally, efficiency is because a large percentage of the MEMs chips is not a mirror, so the whole laser beam isn't reflected, and some power is lost. The scanning approach with a single laser means that there can be a lag between distant pixels, creating minor image distortion and causing problems for developers trying to reconcile with camera images. These are also resolution limited by how fast the mirrors move; even little tiny mirrors have momentum and break when you try to push them too fast.

 

Benefits: The form factor is a significant driver for automotive and other companies: it doesn't spin, so it can be embedded into a grill and hidden. Another benefit is simplicity; MEMS are well-developed, so today's technology is available.

Optical Phased Array

Limitations are beam coherence and precision. The first issue is that sources must be highly calibrated to steer a beam correctly in real-world conditions. Air properties like humidity vary widely; therefore, getting a coherent beam over long distances is challenging.

Eight years ago, Quanergy claimed to do 300m with this technology. After long struggles, they now have managed about 150m but only in mild operating conditions.

A record for any company in real-world conditions, claiming to have 300m, has never been demonstrated).

Second, light has extremely short wavelengths, so to get waves to line up just right, emitters have to be manufactured incredibly precisely. Since this is a CMOS-based approach, this challenge translates to low yields--if anything is messed up on the chip, it has to be thrown out.

Benefits: This approach has no moving parts, which is very reliable. It's also a pure CMOS approach (Lidar on a chip), so it can be made very cheap at scale (< $500 per unit). It also uses a continuous wave approach which translates to less intense bursts of power, and eye safety regulations are not as much of a limitation. This approach is also more efficient and uses a continuous wave approach that enables incoming velocity measurements (Aeva's 4th dimension Cramer loved so much). This last feature is interesting but not game-changing. In scenarios like autonomous driving, you need to know how fast something is coming towards you and how fast it is going in other directions. Think of a truck crossing the road. So they'll end up doing something similar to other approaches where they infer velocity (all components) from multiple measurements.

Digital lidar

Limitations - Having a whole array of lasers pointing in one direction isn't ideal for eye safety regulations, and it limits the power these things can operate. That being said, companies have demonstrated close to 250m range (over the threshold of AVs and other industries), but the range may be restricted compared to other approaches.

 

Benefits - This again has no moving parts and is reliable. It's also a pure CMOS approach but uses pulse instead of continuous sources, so it can be made very cheap at scale (targeted <$100 per unit). It also can have a minimal form factor (this is the LiDAR approach in the new iPhone). Digital does not have a scanning approach like OPA or MEMS but fire multiple lasers at once, so resolution is just limited to how many VSCELs you can fit on a chip (which increases as you shrink the lithography process used to make the chip)

Public companies Technology

Luminar (LAZR)  is currently a mechanical and MEMS solution. The mechanical solution is trusted in the AV industry but is incredibly expensive (what I could find on price varied, but safe to say >$20000). Although mechanical is not as sexy as some solid state technologies, their superior range and large field of view, they'll probably still be a relevant technology for at least another decade or so. However, using their current approach, they will not be competitive on price (Ouster's spinning Lidar has similar performance, but the digital approach inside means 100x fewer parts and cheaper, more scalable manufacturing. Ouster is already projecting 30% YoY price reductions and an ASP of $4300 per unit).

Their MEMs Lidar is supposed to be less expensive ($1000 per unit at scale), but due to the reliability and cost of this approach, I question its long-term relevance. CMOS approaches project significantly lower costs, and unless they can manage significantly different range than digital, I don't see how they can compete. Digital LiDAR isn't resolution limited like MEMS, so companies like Ouster have the potential for higher resolution products at 1/10th the cost. However, MEMS technology is available now (long-range phased array is at least 4-5 years out, Ouster's solid state digital Lidar is 2 years out), so a MEMS solution might gain a couple of years some traction.

 

Innoviz (INVZ) - A MEMS approach, so my previous critiques broadly translate to here as well. They do have a partnership with BMW, and again, this technology is here in the near term and will likely enjoy a few years of very little competition from other solid-state approaches.

Velodyne (VLDR)- A mechanical approach but with even more ridiculous prices; see my first critique of Luminar. Ouster has similar performance at a significantly lower price point. Ouster doesn't put together 64 or 128 individual lasers and sensors like Velodyne. They put them all on a semiconductor, so manufacturing costs are much lower will keep getting lower with scale).

Aeva (AEVA) - Currently a MEMS approach with a frequency modulated continuous wave source, but targeting an optical phased array approach by 2024. Here's the problem. They don't plan on producing the current MEMS solution at scale since its bulky and the continuous wave laser source alone is very expensive. If it works at 300m and they accomplish significant yields, as they say, the optical phased array approach can be a game-changer; the problem is that nobody has been able to do it, and companies like Quanergy (QNGY) have tried for eight years and largely failed. When these guys say a four-year out deadline, they mean they have a rough plan, probably have some promising prototypes, and if everything goes right, they might be able to pull it off. The taping out process for making a new chip takes about two years, so possibly they're projecting that in 2 years, they'll have the technology ironed out enough to tape things out, but I am doubtful. I think there's a significant possibility within the next decade that this technology becomes a reality, but I doubt it will follow Aeva's timeline; it hasn't for multiple other companies that have tried.

 

Ouster (OUST)- a mechanical play currently (but with digital LiDAR on the inside), with a solid-state solution coming 2022. Ouster's digital LiDAR approach enables them to cut costs by a projected 30% YoY, with a targeted $4300 ASP this year for their mechanical LiDARS (which is VERY attractive compared to peers). It also gives them a very modular design; want to change resolution or range for a different tier product? Swap out a chip. Ouster offers over 50 different configurations tailored to different markets and has allowed Ouster to diversify massively (revenue is divided nearly equally over five different industries). Finally, Ouster's digital LiDAR performance will increase naturally, just by moving to a minor lithography process (1st iteration was 120nm, 2nd 40nm, a new chip comes out this year on a more minor but so far undisclosed process). That's right; they are not only getting cheaper as they scale but better. Their solid-state digital LiDAR is poised to do very well, and unlike Aeva, they are already in the process of taping it out. Their current products are almost identical (one column instead of multiple columns of lasers), so there is much lower risk that they can deliver the product on time.

 

[Dietl]

Bloomberg Article about Quanergy

Quanergy Systems Inc. found itself in the center of a sudden frenzy over self-driving cars in 2014. It makes lidar, which bounces lasers off objects to help autonomous cars know what’s nearby. That September, the fledgling company announced a partnership with Mercedes-Benz that would put its devices on cars the automaker was using to test autonomous driving features. The deal established an enviable partnership with one of the world’s most prominent auto brands. In January 2015, the two companies showed off a Mercedes E350 sedan outfitted with Quanergy’s lidar devices at the Consumer Electronics Show in Las Vegas.

At the time, the lidar industry was dominated by Velodyne Lidar Inc., which provided the bulky, expensive sensors Google was using on its autonomous cars. Quanergy made the kind of promise Silicon Valley was built on: it’d shrink existing hardware through science, then sell it at a fraction of the price. Quanergy led the pack, as investors poured money into companies describing new techniques for lidar devices. It has raised $160 million to date at a peak valuation of more than $1.5 billion. Last fall, Quanergy began talking to banks about a potential IPO, setting it up to be one of the first public companies to emerge from the wave of firms making tech for autonomous vehicles.

This July, Daimler AG, the parent company of Mercedes, made another announcement. Daimler said it was running a test program for autonomous vehicles on urban streets in the Bay Area, which included a handful of partners, none of which was Quanergy. For lidar, the robo-taxis would use Velodyne.  Daimler declined to comment on its relationship with Quanergy.

It was a troubling sign for the lidar industry’s first unicorn—and it wasn’t the only one. Quanergy has struggled to deliver products along the timelines it has set out for itself, and has shipped devices that don’t work as well as advertised. Numerous employees have left over the last 18 months, including several at key positions. But Quanergy’s biggest challenge is that its autonomous car business hasn’t developed the way it thought it would. The company has increasingly focused on other applications for lidar, including a plan to help build a digital border wall along the Mexican border, a project that has left some employees ill at ease.

Quanergy has stopped talking about an IPO and has been pursuing new investments in recent months. It has had talks about finding a buyer, according to people with knowledge of the situation. Quanergy backers Samsung Ventures and Sensata Technologies Holding Plc, an auto sensor maker, have expressed disillusionment with the startup, according to people familiar with those firms. Alexia Taxiarchos, a Sensata spokeswoman, said the company remained excited about Quanergy. “Developing these lidar sensors involves solving difficult technical issues that take time to overcome,” she added. In an email, a spokeswoman for Samsung’s investment managers said the firm believed Quanergy was ahead of its competitors making lidar for cars, and had shown meaningful results in the security sector. She said Samsung was aware that Quanergy was about to close a new funding round. In an email, a spokeswoman for Quanergy said it “unequivocally denied” that it was seeking a buyer.

In August, Louay Eldada, the company’s co-founder and chief executive officer, invited Bloomberg News to visit his headquarters and assembly line, which he says is capable of shipping a million lidar devices a year. In a sweeping interview, Eldada admitted faults with earlier products that have since been corrected, cast some criticism as broadsides from competitors and insisted that automotive partners remain optimistic about Quanergy. “People understand we are the real deal,” he said. “We are the veterans.” The company also demonstrated its latest technology to a reporter, although only in controlled conditions that didn’t replicate the needs of automotive companies.

Eldada denied that there were problems with Quanergy’s relationship with its investors or with Daimler, and added that Quanergy was “one of the finalists” to provide the carmaker with lidar going forward. Quanergy has more than 5,000 customers, according to Eldada. He said he couldn’t name them for competitive reasons.

“The hype is falling. ...The closer you get to implementation, the clearer it is how difficult it will be to commercialize.”

Quanergy is seeking new capital, and Eldada said about a third of the investment would come from China. He declined to name the potential investors. Asked if Samsung and Sensata would participate in a new funding round, he said, they had “grabbed all they intended to grab."

Bloomberg also spoke to a half-dozen former employees, all of whom asked to remain anonymous, most of them citing the fear of retaliation. They said execution was a consistent problem at Quanergy. Several former employees described Eldada as a combustible and intimidating presence, the stymying debate about product development and seeing any disagreement as intolerable dissent. When asked about this, Eldada pointed to a framed placard listing the firm’s core values, which include teamwork, transparency, intellectual honesty, and empowerment. “You will find that the level of happiness is high,” he said.

Bloomberg News also spoke to current and former business partners, as well as more than a dozen people in the industry, who also described signs of trouble at the company. Most asked to remain anonymous to speak frankly about sensitive business arrangements.

Quanergy’s critics said the distance between its rhetoric and reality have widened. One former employee said he never saw a single device come off the line at Quanergy that met all of its stated specifications, an allegation the company denies. Morale has been flagging for well over a year, said several former employees who are in contact with current staff members. In July, Ryan Field, a senior employee who was in charge of Quanergy’s development of a chip designed specifically for lidar, told the company he was leaving. Eldada described Field’s departure as a blow, but said it was amicable. Field declined to comment.

In the meantime, Quanergy is pursuing other markets. Eldada has become set on building sensors for an electronic barrier along the U.S.-Mexican border, which he sees as a humane alternative to the border wall being pushed by President Donald Trump.

“People weren’t fixing bugs. ...They were just shipping out stuff that was busted.”

The challenges facing Quanergy point to a broader reckoning looming for the autonomous vehicle industry. Investors have poured about $750 million into lidar companies since 2013, peaking last year at $350 million, according to CB Insights, a firm that tracks venture capital activity. Companies working on automotive mapping, software, and other technologies have also raised money from investors determined not to miss a fundamental revolution in transportation.  But a full-scale autonomous driving industry still seems years away, making business prospects questionable. Startups that attracted interest based on the promise of future breakthroughs are struggling to solve difficult scientific and commercial problems.

“The hype is falling,” said Mike Ramsey, a transportation and mobility analyst at Gartner. “The closer you get to implementation, the clearer it is how difficult it will be to commercialize.” Ramsey said some self-driving tech companies are lowering their expectations, pivoting from building their own software to consulting, and expects to see significant consolidation among the 50 or so companies currently making lidar. “There are not enough musical chairs out there for everyone,” he said. 

Eldada is an intense, confident man whose preferred wardrobe—a black suit with an open-collared white oxford shirt—is at least one step more formal than typical for dressed-down Silicon Valley. When he arrived at an interview with a Bloomberg News reporter, he wore a pin honoring Quanergy’s “Best of Innovation” award from 2017’s Consumer Electronics Show on his lapel.

Eldada traces the idea for Quanergy back to his doctoral research at Columbia University. After graduating, Eldada went on to work at Honeywell and DuPont Photonics, a manufacturing specialist, then a handful of telecom and energy firms. During this period, Eldada said, he began working with technical experts that would make up Quanergy’s core founding team. “Together we developed unique capabilities, designed tools you cannot buy, designed rules they do not teach you in school,” said Eldada.

Jim Disanto was one of first to give Quanergy money. Disanto had just started his own investment fund when he met Angus Pacala, a Quanergy co-founder, in 2012. Eldada pitched Disanto, and the investor’s ears perked up at his description of a specialized computer chip that would contain the entire lidar system. “There’s a race going on,” Disanto remembered thinking. “We’ve got to get into this lidar stuff.”

Lidar sensors work by shooting lasers into the world, then determining the position of objects by measuring how long each beam takes to bounce back. It is widely seen as the most promising technique for autonomous vehicles to sense the world around them. But it also has its unique challenges. Lasers must fan out in various directions, extending a sensor’s field of view beyond a single beam. Velodyne does this by putting its sensors on motorized turntables, shooting lasers in a 360 arc, a technique known as mechanical lidar. Quanergy and other upstarts promised to develop novel ways to steer the radar beams with fewer moving parts, simultaneously making the devices smaller, more durable and cheaper to make. Devices without any moving parts, like the ones Quanergy was developing, were known as solid-state lidar.

Eldada was a fantastic evangelist for this idea. A person who saw his presentation while working at an auto company in 2013 called it “one of the best pitches I’ve ever heard.” In a pitch deck from the same period viewed by Bloomberg News, Quanergy said it would sell to manufacturers of self-driving cars, to dockyard operators looking to automate, and even military contractors searching caves for terrorists. The company was also planning to construct an automated manufacturing facility and design its own specialized chips. Quanergy projected sales to grow from 100 units in 2013 to over 200,000 by 2017.

At the time, Quanergy was selling spinning mechanical devices that worked much like Velodyne’s. But Eldada said his company would have a solid-state lidar by the beginning of 2016. And he delivered, sort of.

Quanergy introduced the S3, its solid-state lidar product, at CES in 2016. One person who received a demo was Evan Ackerman, a contributing editor for the website of the Institute for Electrical and Electronics Engineers. “We can’t tell you much about the demo (it’s under NDA), but we can certainly say that the S3 definitely works as a solid-state LIDAR,” Ackerman wrote in an article at the time. “However, we’re obligated to point out that Quanergy has not yet demonstrated a version of the S3 that performs to the specifications that they announced at their press conference.”

Eldada acknowledged this was true. “Back then it didn’t meet spec,” he said. “That’s absolutely correct.” He dismissed it as run-of-the-mill CES puffery. But there were other signs that things weren’t going according to plan. Months before the 2016 release, Pacala suddenly left Quanergy and later started a competitor, Ouster. (The name could be interpreted as a sly reference to Pacala’s departure, but Ouster’s website says it actually refers to a group of characters in a series of sci-fi novels.) Eldada diminished Pacala’s importance to Quanergy. “He’s a mechanical engineer, and that’s all he does and all he knows. He’s never had a hand in the solid-state lidar,” Eldada said—even though Pacala shares credit on a patent entitled “optical phased array lidar system” that describes Quanergy’s approach to solid-state lidar. Pacala declined to comment.

Employees who were at Quanergy at the time said Pacala’s departure shook morale, especially after Eldada began criticizing him and his new company in front of other staff members. They saw Eldada’s continued reproach of Pacala as indicative of the CEO’s hunger for personal loyalty. This made it hard to fix technical issues, said one person who held a senior position at the company. “You expect a certain number of failures when you’re building a business, because it’s new stuff, and nothing is perfect. But when it goes on for a while, you think that is a problem,” this person said. “And when you combine that with an environment where passionate debate is not tolerated, it’s difficult for you to step in and say, ‘Why are we having this problem?’”

Quanergy’s own public statements about its technology hint at technical frustrations. It first described the potential of its mechanical lidar device, the M8, in a December 2014 press release, when it said it had a range of 300 meters. By January 2017, the company said the device had a 200-meter range. Six months later, it revised its claim again, saying it had a “maximum range exceeding 150 meters.” By November 2017, it issued a press release describing its range simply as “long.” 

In 2016, researchers at PSL Research University in Paris tested an M8 device themselves, and found the error rates Quanergy said its device would experience at 50 meters actually happened at 11 meters, when tested outside. They also found it performed far worse outdoors than indoors, a red flag for car makers. When asked about the study, former Quanergy employees said it was common knowledge within the company that performance varied drastically for each device it shipped, a symptom, they said, of the lack of firm production standards. According to two former employees, customers returned M8 devices at alarming rates. “People weren’t fixing bugs,” one said. “They were just shipping out stuff that was busted.” In an email, a spokeswoman denied this. “Quanergy has never knowingly shipped product with defects,” she said. She added that no clients have returned the current version of the M8, released in February.

Eldada said the stated specs in press releases vary because they refer to different versions of the product. He didn’t dispute the PSL study, but said it was out of date. “I’m sure that’s what they saw,” he said. “We’re actually going back to the customers who tried our product three years ago and saying, ‘Forget about it. It’s a totally different product,” he said. Eldada said he created three different versions of the M8 this year and that 94 percent of the sensors now have 200 meter range. In any case, the future of the company, in Eldada’s eyes, has always been the S3.

For now, the officially stated specifications of Quanergy’s S3 sensors fall short of the performance needed for general-use autonomous driving. A commonly cited benchmark for the needs of highway driving is a 200-meter range, which would allow a car driving at 70 miles per hour just over 6 seconds to adjust to an obstacle. A current spec sheet viewed by Bloomberg News shows a range of 150 meters.  Quanergy also measured its range on highly reflective objects, which are easier for lidar to see at long distances but aren't usually the type of things autonomous cars need to spot. This elides how well the S3 might sense many of the objects autonomous cars would seek to avoid.

There’s a consensus within the industry that no one has yet figured out the perfect formula. “There’s been some delay, the technology in general is difficult,” said Eugene Zhang, a partner at TSVC, another Quanergy investor. “I believe in the team. They have a good shot at it.”

Disanto, the early investor, said that the company’s new devices remain a work in progress, but that he believes the company is ahead of the rest of the industry. “I’ve been to the company. I’ve seen the new samples,” he said. “We’re really close.”

One auto company that has expressed interest in the S3 is Japan’s Koito Manufacturing Co. In 2017, Quanergy announced a collaboration with the firm to make headlights with built-in lidar sensors. But Daisuke Sato, a Koito spokesman, said this month that the deal wasn’t final. “We still haven’t decided which suppliers we should buy parts from,” he said.

Eldada expressed confidence that he could win Koito over. He said Quanergy’s devices could outperform their stated specs in certain conditions such as highway driving, when sensors need to see at longer ranges. He said Quanergy’s device was on par with competing products. In a demo conducted for Bloomberg News, Eldada showed an S3 operating in a conference room that was about 12 meters long. When asked to see the device work outdoors he declined, citing a firmware update.

Quanergy has consistently said its tech would be important for more than just cars. In 2016, the company floated the idea of using lidar to secure the perimeters of prisons, and later engaged with oil companies, on plans to use its sensors to detect intruders.

People throughout the industry have noticed a distinct shift toward these kinds of applications over the last 18 months. The most notable project is Quanergy’s attempt to build technology for a digital border wall. In Eldada’s vision, lidar sensors could be deployed along the border, monitoring remote regions for movement, then determining whether there were people attempting unauthorized crossings. Two former employees said many workers found the idea repellant.

Eldada acknowledged that a contract for a virtual wall would be controversial, but he described it as a kind of protest against Trump’s version of border security. “I hate the word ‘wall’. We are anti-wall. We have a technology solution,” he said. A digital border project could be lucrative, but could also help open up other markets, like selling technology to cities or airports, said Eldada. It would also require addressing challenges that don’t arise with cars, like operating in extreme heat without consistent access to power or data networks. Quanergy sells sensors to Cisco for urban planning, but Quanergy declined to elaborate on the deal.

In an email, a spokeswoman for Quanergy denied that the company is primarily focused on the security market. But none of the deals that the company has announced this year are with auto companies. “We don’t want to put all of our eggs in one basket, and be subject to delays that no one can control,” said Eldada. Not that he’s giving up. “Ultimately, when it takes off in a big way, it will dwarf everything else,” he said. 

—With assistance from Gabrielle Coppola, Alex Barinka, and Masatsugu Horie

[Dietl]

LiDAR Overview
To begin to address this question, it is necessary to understand the anatomy of a LiDAR system, of which there are different architectures. Coherent LiDAR, a type of which is referred to as frequency-modulated continuous wave (FMCW), mixes a transmitted laser signal with reflected light to compute the range and velocity of objects. FMCW offers some advantages but it remains relatively uncommon when compared to the most common LiDAR approach, direct time-of-flight (dToF) LiDAR. This implementation measures distance to an object by timing how long it takes for a very short pulse of light sent out from an illumination source to be reflected off an object and returned to be detected by the sensor. It uses the speed of light to directly calculate the distance to the object using the simple mathematical formula relating time, speed, and distance. A typical dToF LiDAR system has six major hardware functions, although the choice of wavelength mostly impacts the transmit and receive functions.

Figure 2: A block diagram of a typical dToF system with green portions representing some focus areas of ON Semiconductor products.

Table 1 shows a list of the various LiDAR manufacturers that range from known automotive Tier-1s to startups across all regions of the globe. Based on market reports and public information, the vast majority of these companies operate their LiDARs at near-infrared (NIR) wavelengths, as opposed to short wave infrared (SWIR) wavelengths. Furthermore, while the SWIR-focused suppliers working on FMCW are restricted to those wavelengths, most of those with a direct time-of-flight implementation have a path to making a system with NIR wavelengths, should they choose, while being able to leverage a lot of their existing IP around functions such as beam-steering and signal processing. 

Table 1: List of LiDAR manufacturers that operate in NIR and SWIR wavelengths. Not a comprehensive list. (Image source: Yole, IHS Markit, and public disclosures)

Given that the majority, but not all, of these manufacturers have chosen NIR wavelengths, how they came to this decision and what the implications are should be considered. At the heart of the discussion is some basic physics related to the properties of light and semiconductor materials making up the components used in LiDAR. 

Photons fired by the laser in a LiDAR system, which are intended to be bounced off objects and received by the detector, have to compete with ambient photons coming from the sun. Looking at the spectrum of solar radiation and taking into account atmospheric absorption, there are “dips” in the irradiance at certain wavelengths that would reduce the amount of photons existing as noise for the system. At 905nm, there is about 3x higher the amount of solar irradiance than at 1550nm, meaning a NIR system has to contend with more noise that can interfere with the sensor. But this is just one of the factors to take into account when choosing a wavelength for a LiDAR system.

Figure 3: Atmospheric absorption of light results in clear peaks.

Sensors
The components responsible for sensing the photons in the LiDAR system are different types of photodetectors, so it is important to explain why they may be made up of different semiconductor materials depending on the wavelength to be detected. In a semiconductor, a band gap separates the valence and conduction bands. Photons provide the energy to help electrons overcome that band gap and make the semiconductor conductive, thus creating a photocurrent. Every photon’s energy is related to its wavelength, and a semiconductor’s band gap is related to its sensitivity — this is why different semiconductor materials are needed depending on the wavelength of light that is to be detected. Silicon, which is the most common and cheapest semiconductor to manufacture, is responsive to visible and NIR wavelengths up to about 1000nm. To detect wavelengths beyond that in the SWIR range, alloying of more exotic group III/V semiconductors can be done to make materials like InGaAs capable of detecting those wavelengths of light, from 1000nm to 2500nm. 

Early LiDARs used PIN photodiodes as sensors. PIN photodiodes have no inherent gain and as a result, are not able to detect weak signals easily. Avalanche photodiodes (APDs) are the most prominent type of sensor used in LiDAR today and provide a moderate amount of gain. However, APDs also need to operate in linear mode like PIN photodiodes to integrate signal from photon arrivals and also suffer from poor part to part uniformity, while requiring very high bias voltages. The newest types of sensors that are increasingly being used in LiDARs are built on single photon avalanche diodes (SPADs), which have a very large gain and are able to produce a measurable current output from every single photon detected. Silicon photomultipliers (SiPMs) are arrays of silicon-based SPADs that come with the added advantage of being able to distinguish single photons from multiple photons by looking at the amplitude of the generated signal.

Figure 4: Different types of photodetectors used to detect signals in a LiDAR

Circling back to the relevance to the topic of wavelengths, all of these types of photodetectors can be built on silicon (for NIR detection) or III/V semiconductors (for SWIR detection). On the other hand, manufacturability and cost are key to viability for the technology, and CMOS silicon foundries allow for high-volume and low-cost manufacturing of such sensors. This is a primary reason why SiPMs are being increasingly adopted for LiDAR on top of allowing for higher performance. While APDs and SPADs for SWIR exist, it is difficult to integrate them with readout logic due to the fact that the processes are not silicon-based. Lastly, III/V-based SPAD arrays and photomultipliers (analogous to SiPMs) for SWIR have not yet been commercialized, so the ecosystem availability favors the NIR wavelengths.

Lasers
Generating photons involves an entirely different process. A semiconductor p-n junction as the gain medium can be used to make a laser; this is done by way of pumping a current through the junction causing the resonant emission of photons as atoms go to lower energy bands, resulting in a coherent laser beam output. Semiconductor lasers are based on direct band gap materials like GaAs and InP, which are efficient for the generation of photons that happens when atoms go to a lower energy band, unlike indirect band gap materials such as silicon.

There are two main types of lasers used in LiDAR: edge-emitting laser (EEL) and vertical cavity surface-emitting laser (VCSEL). EELs are more widely used today, owing to their lower cost and higher output efficiency than VCSELs. They are more difficult to package and build into arrays and also suffer from a wavelength shift across temperature which causes the detectors to have to look for a wider band of photon wavelengths, allowing for more ambient photons as noise to also be detected. Despite the higher cost and lower power efficiency, the newer VCSEL technology has the advantage of easy and efficient packaging since the beam is generated from the top. The market adoption of VCSEL is increasing as its costs will continue to decrease significantly and the power efficiency will improve. EELs and VCSELs exist for both NIR and SWIR wavelength generation, with a key difference — NIR wavelengths can be generated with GaAs, while SWIR wavelengths require the use of InGaAsP. GaAs lasers are able to use larger wafer size foundries leading to lower cost, again pointing to an advantage for the ecosystem of NIR LiDAR manufacturers from both a cost and supply chain security perspective.

Figure 5: Different types of lasers used in a LiDAR.

Laser Power and Eye Safety
While talking about the wavelength debate, it is imperative to consider the eye safety implications of a LiDAR system. The concept of dToF LiDAR involves using short laser pulses with a high peak power over a certain angle of view to be emitted to the scene. A pedestrian standing in the path of a LiDAR’s emission path needs to be assured that their eyes will not be damaged by a laser being fired in their direction, and IEC-60825 is a specification that dictates how much the maximum permissible exposure is across the different wavelengths of light. While NIR light, similar to visible light, is able to pass through the cornea and reach the retina in the human eye, SWIR light is mostly absorbed within the cornea, and as a result, is able to be exposed at higher levels. 

Figure 6: IEC-60825 specification for eye-safe laser exposure.

Being able to output multiple orders of magnitude higher laser power is an advantage for a 1550nm-based system from a performance perspective, as it allows for more photons to be sent out and thus be returned to be detected. Higher laser powers also come with a thermal tradeoff though. It should be noted that proper eye-safe design has to be done regardless of wavelength by clearly taking into account the energy per pulse and the size of the laser aperture. With a 905nm-based LiDAR, the peak power can be increased by either of these factors, as shown in Figure 7.

Figure 7: Eye-safe laser design for a NIR LiDAR given different optics and laser parameters.

Comparison of NIR and SWIR LiDAR Systems
The above focus on the amount of laser power able to be output brings us back to the sensors being used. A higher-performance sensor that is able to detect weaker signals will clearly benefit the system in multiple ways — in being able to achieve longer range or being able to use less laser power to achieve the same range. ON Semiconductor has developed a series of SiPMs for NIR LiDAR driving the photon detection efficiency (PDE) — a key parameter indicating sensitivity — to a market-leading 18% with its latest RDM-Series sensors.

Figure 8: Process roadmap of ON Semiconductor SiPMs.

To compare the performance of a NIR dToF LiDAR with a SWIR dToF LiDAR, we performed system modeling for identical LiDAR architectures and environmental conditions with differing parameters for the lasers and sensors. The LiDAR architecture is a coaxial system with a 16-channel detector array and a scanning mechanism to spread across the field of view, as shown in Figure 10. This system model has been validated with hardware and allows us to accurately estimate the performance of LiDAR systems.

Figure 9: System model for a dToF LiDAR sensor.

Table 2: LiDAR sensor and laser parameters for NIR and SWIR system model simulation.

The 1550nm system uses a higher amount of laser power, as well as a higher PDE sensor owing to its use of high-PDE InGaAs alloys, which should yield better-ranging performance in our system simulation. Using system-level parameters of 100klux ambient light filtered by a 50nm bandpass on the sensor lens (centered around 905nm and 1550nm respectively), a 0.1° x 5° angle of view scanned over 80° horizontally at 30fps, a 500kHz laser repetition rate with 1ns pulse width, and a 22mm lens diameter, the results are shown in Figure 10.

Figure 10: Simulation results for similar LiDAR systems based on 905nm and 1550nm.

As expected, the 1550nm system is able to range further for a low-reflectivity object, going up to 500m with 99% ranging probability. However, the 905nm-based system still achieves well over 200m of ranging, showing both types of systems can achieve automotive long-range LiDAR requirements in typical environmental conditions. When poor environmental conditions like rain or fog are introduced, the water absorption properties of SWIR light cause its performance to degrade more rapidly than a NIR-based system, which is another factor to take into account.

Cost Considerations
Having looked extensively at the technology behind LiDAR systems, as well as the implications of using different wavelengths, we now go back to the cost considerations factor. We earlier explained that the sensors being used for NIR-based LiDARs come from native CMOS silicon foundry processes, which enable the lowest possible cost for semiconductors. In addition, they also enable integration of CMOS readout logic with the sensor into one chip by use of stacked die technology, which is readily available at foundries today, further collapsing the signal chain and reducing cost. Conversely, SWIR sensors use III/V semiconductor foundries like InGaAs which are higher cost and new hybrid Ge-Si technology — which may enable lower-cost SWIR sensors — making integration with readout logic easier but are still estimated to be more than 5x more expensive than traditional CMOS silicon even after reaching maturity. On the laser side, the difference in wafer size between the GaAs wafers used for making the laser chips in NIR systems versus the InGaAs wafers used for making the laser chips in SWIR systems again leads to a cost disparity, and the fact that NIR systems have a path to using VCSELs with a much more readily-available supplier base also enables lower-cost integration.

The sum total of the above factors led to an analysis done by IHS Markit (Amsrud, 2019), which showed that for the same type of component (the sensor or laser), the cost for a SWIR system would be 10 to 100 times higher than a NIR system. The average combined component cost for the sensor and laser for a NIR system was estimated to be between $4 to $20 per channel in 2019 and decreasing to $2 to $10 by 2025. By contrast, the equivalent average component cost for a SWIR system was estimated to be $275 per channel in 2019 and decreasing to $155 per channel by 2025. That is a tremendous cost difference when considering the fact that LiDAR systems contain multiple channels, even if using a 1D-scanning approach since a vertical array of single point channels is still required.

Table 3: Summary of cost considerations. (Image source: IHS Markit)

The LiDAR market dynamics also do not favor the SWIR camp. The autonomous driving market has not ramped as quickly as market expectations five years ago, and Level 4 and Level 5 autonomy systems, for which LiDAR is a must, are years away from widespread mass deployment. In the meantime, the industrial and robotics markets making use of LiDAR are even more cost-conscious and have no need for the ultra-high-performance advantages of a SWIR system, so these manufacturers do not have a way in the meantime to bring component costs down by ramping volume as is often claimed. There is a “chicken and the egg” problem of getting the lower cost when the volume ramps but needing the lower cost to get the volumes.

Summary
After doing a deep dive into the technology and the differences between NIR and SWIR systems, it is clear why the vast majority of LiDAR systems in existence today are using NIR wavelengths. While the outlook for the future is never 100% certain, it is apparent that the cost and availability of ecosystem suppliers are key factors, and NIR-based systems will certainly always be cheaper due to the technology advantage and economies of scale for CMOS silicon. And while SWIR does allow for a longer-ranging LiDAR system, NIR-based LiDARs can also achieve desired automotive long-range requirements, while also performing extremely well for short- to medium-range configurations also needed in ADAS and AD. The existence of NIR-based LiDARs in mass production for the automotive market today shows that the technology has been commercialized and proven out, but it will still take some time for consolidation to happen and for the winners and losers to shake out. After all, the automobile industry at the turn of the 20th century contained 30 different manufacturers, and that increased to nearly 500 over the next ten years — but it only took a few years after that for most of them to disappear. It is expected that a similar dynamic may happen with LiDAR manufacturers by the end of this decade.

https://www.eetimes.com/demystifying-lidar-an-in-depth-guide-to-the-great-wavelength-debate/

[Dietl]

Sense Photonics’s direct Time-of-Flight platform delivers a unique value proposition to the automotive industry:  a lidar platform which does not compromise performance for price. To make sense of the wide array of lidar technologies in the market today, it helps to divide the market into 4 generations of lidar systems:

1st Generation: Mechanically-Rotating Linear Scanning Systems

Mechanically-rotating lidars integrate multiple discrete lasers and detectors in a single enclosure. Vertical resolution is defined by the number of lasers which are packed in the enclosure while a 360 degree field-of-view is attained by fast mechanical rotation of the whole system. 

While this was an exciting technology at the outset of lidar, and helped launch the technology during the DARPA Grand Challenge competition in the 2000’s,  rotating systems offer limited resolution for a high price, because their resolution is defined by the number of discrete lasers which can be assembled into a small enclosure, and by the fine alignment between those laser and the discrete detectors which collect their signal. 

Moreover, these systems suffer from several additional deficiencies common to scanning systems, such as motion blur, rolling shutter, distortion, and reliability issues caused by vibration or shock.   These systems were designed to optimize system performance but neglected SWAP-C3 considerations (size, weight, power, cost, cooling, and compliance). Specifically, they tend to be large and produce 360 degree fields of view, regardless of whether those are necessary. As such, they typically protrude from the car’s roof and do not lend themselves for aesthetic and aerodynamic integration into the vehicle.

2nd Generation: Micro-Actuated Point Scanning Mirrors

In response to the deficiencies of rotating mirror lidar systems, several companies developed micro-actuated scanning systems. These systems operate by deflecting one or more collimated laser beams as well as the received signal from one or more mirrors, thus scanning the field-of-view and directing target echos to a detector. Beam steering is achieved either using discrete microactuators or using an integrated MEMS chip. 

Micro-actuated scanning systems suffer from a number of critical deficiencies. Most notably, these systems trade-off resolution for field-of-view when operating at a given frame rate.

For example, take such a point-scanning system operating at 25 frames-per-second (or 40 msec frame time). Since light takes 1.63 μsec to travel to a 250 m target and back, a singlebeam can only capture 40 msec/1.63 μsec = 24,500 target-directions per frame. If the system’s angular resolution is 0.05° x 0.05°, then the total field of view would be limited to about 60 square degrees, for example 7.7° x 7.7°. Such a narrow field of view is typically insufficient for long-range applications, forcing such lidar systems to operate at variable resolutions across the field of view. Variable resolution runs the risk of missing unexpected objects and creates a computational burden on downstream data fusion engines. 

Similar to Gen 1 systems, Gen 2 systems also suffer from motion artifacts. For example, since Field-of-View acquisition is sequential, an object may be imaged for a first time when the vehicle isin a first position and then the same object may be scanned again when it appears at a different direction due to the vehicle’s changing position. Thus, it is not uncommon to see a street sign appear twice in a point cloud with such systems.

Micro-actuated systems also suffer from mechanical deficiencies. Due to the tight tolerancing required to achieve their angular resolution, these systems are more susceptible to vibration, shock and temperature effects. They also need to be carefully aligned, which, among other factors, contributes to their high cost.

Whether utilizing direct Time-of-Flight signal acquisition or FMCW (Doppler-shift), current Gen 2 lidar systems continue to struggle with complexity, reliability as well as motion artifacts.

3rd Generation: Electronically-Scanning Lidar

The quest for more reliable and cost-effective automotive lidar led some companies to develop electronically scanning lidars. These systems utilize electronics to steer their laser beam without mechanical motion. These methods include Optical Phase Array (OPA) as well as utilizing VCSEL arrays with integrated microlenses which direct the light output of portions of the array to various subregions of the field-of-view. While Gen 3 lidars offer higher reliability than mechanically scanning lidars, they still suffer from motion artifacts because of their sequential acquisition of the complete point cloud. Moreover, a number of the scanning technologies have proven to lack maturity or to be sensitive to temperature or to device-to-device variabilities.   

4th Generation: Solid-State Global Flash Lidar

The holy grail of automotive lidar is a mass-manufacturable, cost-effective and high-resolution lidar, which acquires point clouds without any scanning.  Numerous technological challenges have prevented developers from creating such non-scanning flash automotive lidar systems. High peak optical power, uniformly illuminating a large field of view, has traditionally translated to large power and expensive systems with eye safety limitations. On the receiver end, non-scanning systems require high-resolution focal plane arrays with simultaneous parallel signal processing.

Sense Photonics has developed the only flash lidar system meeting market requirements. The company builds its custom emitter by printing a large number of Vertical-Cavity Surface-Emitting Lasers (VCSELs) using its proprietary Micro Transfer Printing process. Eye safety and reliability are assured by diffusing the outputs of the VCSELs to illuminate the full field of view. For its receiver, Sense selected CMOS Single-Photon Avalanche Diodes (SPADs) integrated in an internally-designed smart imager sensor chip, which processes more than 100,000 pixels’ histograms simultaneously. The integrated system is now shipping to select customers, delivering unprecedented point cloud resolutions across wide fields of view.

Sense’s 4th Generation lidar utilizes mass-manufacturable and automotive-qualified technologies, achieving unprecedented cost efficiencies. Its unique architecture results in true global shutter acquisition, thus removing motion artifacts which are prevalent in legacy lidar generations. Since no scanning is required, the system is robust to vibration and shock, and does not require fine alignment or calibration. With its camera-like architecture, these systems can be efficiently integrated into vehicles in an aesthetic and aerodynamic way.

 

Conclusion

The arrival of true solid-state global flash lidar will allow, for the first time in the industry, lidar systems which do not compromise cost for performance.  Sense is the only flash lidar company in the world which has achieved high-precision, 200m range using high-volume-manufacturable components. This advancement will allow OEMs to optimize SWAP-C3 parameters while still delivering exponential improvements in Points-per-Second over the three legacy lidar generations.  Combining the scalability and manufacturability of VCSELs with solid-state silicon CMOS SPAD technology , Gen 4 lidars deliver a true paradigm shift in the industry enabling the mass adoption of autonomous and assisted driving systems.

[Dietl]

Aeva update, no revenue forecast for 2022 

[Dietl]

After MVIS reports

 

 

[Dietl]

Updates with Velodyne and Luminar. Velodyne decided not to provide 2022 forecast, I put it slightly over the mark for 2021

 

[Dietl]

Income statement comparison 

 

[Dietl]

Updated with Innoviz results

 

[Dietl]

Update for INVZ, expected revenue $10M not disclosed by the company. 

 

[Dietl]

Advancements in Diode Lasers Fuel Automotive Lidar

    

 

Semiconductor lasers coupled with beam-shaping optics are the engine driving continued development of more capable, compact, and cost-effective automotive lidar systems.
 

LEON LI, FOCUSLIGHT TECHNOLOGIES INC.

Lidar for autonomous vehicles debuted in 2007 as an enabling sensor technology for robotaxis and robotruck development. Within the last five years, however, the technology has trended toward broader commercialization in mass-market passenger vehicles, as evidenced by a dozen volume contracts or project nominations signed by OEM carmakers and lidar providers. Mercedez-Benz, for example, announced in December 2021 that its first model to offer Level 3 autonomy will be equipped with lidar. The number of contracts will grow even faster in the next few years as lidar enables more passenger vehicles with Level 2 or Level 3 autonomy.

The technology’s accelerating commercialization underscores how far lidar has evolved beyond its earlier architecture, in which dozens of laser emitters and avalanched photodiodes were assembled onto a 360º rotating base. A number of innovative lidar architectures and sensor technologies have since been developed. Among the various architectures’ popular scanning methods are MEMS-based laser-spot 2D scanning, rotating mirror line-beam 1D steering, mixed 2D mechanical scanning, and flash illumination without any scanning.
   

---

Line-beam-shaping optics for line-beam-steering lidar applications. Courtesy of Focuslight Technologies.

---

A unifying theme underlies this diver­sity of developing lidar technologies, which is that the continued commercialization of the technology calls for further development in laser technologies. More specifically, this call applies to the four laser types commonly used for automotive lidar systems. They are edge-emitting lasers (EELs), vertical-cavity surface-emitting lasers (VCSELs), diode-pumped solid-state lasers (DPSSLs), and pulsed fiber lasers.

Solid-state and fiber lasers offer relatively higher pulsed energy, eye-safe emissions in the shortwave IR range, and higher beam quality versus EELs and VCSELs. But the latter two laser technol­ogies also have their advantages.

Largely based on gallium arsenide (GaAs) compounds, EELs and VCSELs are semiconductor lasers that operate primarily at NIR wavelengths, such as 905 or 940 nm, and are well known for their superior conversion efficiency, simplicity, and compatibility with automotive standards, and for offering the most compact form factor versus other lidar sources. As a consequence, semiconductor laser technology occupies the highest market share in the automotive lidar market, with EEL taking the predominant share.

VCSELs, however, have become a growing alternative. Widely used in telecom and datacom transceivers, and more recently in consumer electronic devices for 3D sensing, VCSELs are now targeting the automotive lidar market. Manufacturers have developed multijunction VCSEL devices, with the latest devices incorporating up to five or six junctions that generate several-times-higher peak power densities to meet the demand for lidar systems.
 
 

---

Automotive lidar systems typically leverage one of four laser types: edge-emitting lasers (EELs), vertical-cavity surface-emitting lasers (VCSELs), diode-pumped solid-state lasers (DPSSLs), and pulsed fiber lasers. Each has merits and drawbacks (top), and each has been adopted by lidar developers to varying degrees (bottom). Based on statistics compiled by Focuslight in a survey of over 50 lidar developers. Courtesy of Focuslight Technologies.

---

There are well-known pros and cons for VCSELs. While their power density per lasing area is an order of magnitude lower than EELs, they exhibit significantly lower wavelength temperature dependence and are less prone to facet damages, which translates into higher reliability. They allow easier implementation of two-dimensional emitter matrices with individually addressable rows and columns. This last advantage is significant enough to make VCSELs a preferred choice for solid-state flash lidar systems, as well as for their variation — sequential or segmented flash lidar for short- to mid-range detection at wide field of view (FOV). Ultimately, VCSEL technology could potentially become the most cost-effective option for high-volume production due to its efficient wafer-level coating, testing, and screening processes, as well as its compatibility with high-volume production.

Beam shaping

Despite their many merits, both VCSELs and EELs still face several challenges when applied in a lidar system. Semiconductor lasers are as famous for their poor beam quality as they are for their high optical efficiency.

A typical 10- × 200-µm nanostack EEL emitter with 120 W of nominal pulsed power has a typical beam divergence of 25° in the vertical axis and 10° in the horizontal axis, defined as full width half maximum (FWHM). The result is an elliptical beam pattern in the far field. In contrast, a typical VCSEL, regardless of its dimensions, usually has a symmetrical divergence around 25° on both axes, defined as FWHM.
 

 

---

A typical EEL fast- and slow-axis collimator designed for use in MEMS lidar systems (top). Beam divergence for such systems (bottom) is typically 0.1° × 0.8° (middle). Courtesy of Focuslight Technologies.

---

These poor parameters make semiconductor lasers inadequate to be applied directly to mainstream lidar systems. Therefore, it is critical to ensure that they have a proper beam-shaping optical design that directs laser photons appropriately and transforms the original beam shape into the required beam patterns.

One common approach combines fast-axis and slow-axis collimation for the nanostack EELs used in MEMS-based lidar systems.

This combination requires careful design considerations to achieve the best possible collimation on both axes while at the same time restricting the beam size and making sure it is within the MEMS mirror’s clear aperture.

As lidar technology and architectures have evolved, two other advanced beam-shaping techniques have emerged for semiconductor lasers: the line-beam concept and the flash-illumination concept.

A line beam is usually generated by an EEL minibar fabricated as a linear array of nanostack EEL emitters. A line-beam-shaping design usually consists of a long-focal-length aspherical fast-axis collimation lens that generates horizontal divergence as small as <0.1° and a line-beam homogenizer that produces a typical and customizable 25° vertical divergence with high uniformity across its intensity distribution. Coupling these beam- shaping techniques with a mechanical rotating mirror and a novel silicon photomultiplier (SiPM) or single-photon avalanche diode (SPAD) array on the detector end enables a new generation of hybrid solid-state high-resolution beam-steering lidar. To achieve such line-beam-shaping configurations, it is critical to generate a narrow, uniform, and clean-line laser beam with very low divergence in the fast axis, high uniformity, and minimum intensity outside the designed FOV.

The flash-illumination beam-shaping concept often utilizes two directional microlens arrays as diffusers in front of the VCSEL to generate a rectangular FOV with high uniformity or, in other cases, an FOV with specifically defined intensity profiles. Focuslight recently developed ultrawide-angle diffusers that can generate close to 160° FOV with a batwing intensity distribution, enabling ultrawide FOV for lidar or in-cabin sensing applications.

These beam-shaping concepts and solutions are advancing semiconductor laser applications for lidar by lowering system complexity and increasing system signal-to-noise ratios.

Advancements in the works

No perfect laser solution exists for lidar. However, laser and optics manufacturers are accelerating the pace of their innovations and research to address technical challenges and meet the fast-growing demand from the lidar market. Several potential advancements in semiconductor laser technology show promise.

For example, pulsed EELs are often packaged in a special quad flat no-lead (QFN) or transistor-outline (TO)-can configurations for automotive qualification and customer applications. Both laser package types present drawbacks, such as compromised thermal performance or higher parasitic inductance during short-pulse operation. An alternative approach under development is to bond the EEL die directly onto a driver printed circuit board (PCB) or a ceramic substrate for further packaging. EEL bare die bonding technology could provide an improved laser die packaging solution for lidar based on EEL minibars.

Another emerging advancement involves EELs with temperature- dependence characteristics comparable to VCSELs.
 
 

---

As lidar technology and architectures have evolved, advanced beam-shaping techniques have emerged for semiconductor lasers. The flash-illumination beam-shaping concept for VCSELs (top) utilizes two directional microlens arrays as diffusers in front of the VCSEL to generate a rectangular field of view (FOV) with high uniformity (bottom). Ultrawide-angle diffusers are a more recent development that can generate close to 160° FOV with a batwing intensity distribution, enabling wide-FOV lidar or in-cabin sensing applications. Courtesy of Focuslight Technologies.

---

Wavelength-stabilized EELs that share a comparable wavelength shift coefficient to VCSELs (e.g., 0.07 nm/ºC) could reduce the need for thermoelectric coolers for temperature control. They could also help narrow the spectral range of bandpass filters used on a lidar system’s receiver end to improve the system’s signal-to-noise ratio.

Semiconductor laser-makers are also tailoring their devices for lidar applications by adding more junctions to EEL and VCSEL architectures in a race to achieve higher peak powers. EELs have had the edge, so to speak. But VCSEL technology is catching up, with the development of devices incorporating up to eight junctions. EELs with four or five junctions are also being developed and tested in labs. Unlike VCSELs, EELs with more junctions offer an increased active emission area, which increases the challenges for beam shaping. Semiconductor laser designs with more laser junctions are attractive to lidar developers because laser slope efficiency and peak power density both increase proportionally with the number of junctions. Importantly, adding junctions also brings more challenges with regard to thermal design, manufacturing yield, and long-term reliability. Therefore, thorough qualification of these lasers must be completed before they can be used in commercial lidar systems.

Another potential breakthrough involves back-emitting VCSEL designs. Such devices would allow VCSEL production to leverage surface-mount technology, which would reduce the wire bonding and parasitic inductance that is common to VCSEL packaging and allow faster rise and fall times and shorter laser pulses. Back-emitting VCSELs also enable micro-optics to be etched on the GaAs wafers directly. This could be a potential game changer because it could enhance optical performance and significantly reduce lidar system complexity.

Semiconductor laser-makers are also developing devices that emit at eye-safe SWIR wavelengths between 1340 and 1550 nm. NIR wavelengths such as 905 or 940 nm, at which semiconductor lasers typically operate, are only eye safe under a specific upper limit of energy density and eye exposure. In contrast, the 1470- and 1550-nm wavelengths permit an upper limit of energy density that is several times higher. Recent development of indium phosphide (InP)-based EELs with multiple junctions aim to deliver higher peak powers within the SWIR range for automotive lidar applications. The latest devices to emerge exhibit more than 100 W of peak power from a 200-μm strip width, with a three-junction InP EEL lasing at 1550 nm — a significant advancement. Nevertheless, it is still far from the peak powers that can be generated by a 1.5-μm pulsed fiber laser, so it has not yet made semiconductor sources a primary choice for commercialized lidar operating at these wavelengths. Producing high-power VCSELs that operate at SWIR wavelengths has also been consistently more challenging. While some recent cases use SWIR VCSELs as the illumination source for SWIR sensing systems, these sources still do not offer a primary solution for lidar applications.

Road to mass production

A semiconductor laser-based lidar system or its components and subassemblies must pass four milestones to be compatible with automotive-grade mass-production applications.

First, the system must be designed to deliver automotive-grade performance. This means a semiconductor laser and its beam-shaping optics must collectively perform within target specifications over a wide operating temperature range per automotive grade. EEL beam pointing and directivity shifting under wide operating temperatures is one of the top challenges toward reaching this goal. And significant research and development on both opto- and thermal-mechanical designs of these systems is required to minimize the optical performance temperature dependence.

Second, semiconductor laser systems must be tested and qualified as capable of delivering automotive-grade reliability. Despite the readiness of discrete semiconductor lasers and optics, customized automotive-grade qualification of an integrated laser-optics assembly remains a top challenge for most lidar developers.

Automotive reliability qualification plans typically begin with a design verification stage on a smaller sample size of a few dozen laser systems followed by a product-validation step that can last from six to as many as 12 months, based on a few hundred samples. Passing these stringent qualifications for durability under high-temperature operating conditions, for example — or withstanding high-humidity testing — requires careful attention to laser, optical, thermal, and mechanical designs to ensure that their finished assembly will achieve an adequate reliability margin. Robust mounting of optics and reliable adhesive joints are critical to surviving these tests.

The third milestone that laser assemblies confront is the construction of an automotive-grade high-volume manufacturing line with high-level process automation. For critical processes, automotive parts manufacturers apply a process capability index (Cpk) to measure and monitor the ability of a process to meet target specifications. Automotive-grade semiconductor laser assemblies are no different. They must have a Cpk value that meets automotive standards to ensure that continuously produced samples can either pass product verification qualifications with zero failures or be delivered to automotive customers with zero defects. A failure in any qualification test will lead to a failure in the automotive-grade qualification. Achieving a passing Cpk value is only possible by using a high level of process automation, especially for the high production volumes forecast for automotive lidar in the next few years.

The final milestone must put in place a stringent and well-established automotive-grade quality management process following the International Automotive Task Force’s 16949 standard to ensure zero defects for continuous high-volume production and deliveries. This process is nothing new for automotive parts suppliers. But implementing it could be quite a challenge for most laser and optics manufacturers serving industrial markets with low to moderate production volumes.

Scanning forward

As any laser or optical technology matures and moves closer to mass production, the final and ultimate challenge will involve cost. Lidar developers targeting the automotive market face cost challenges that have never been seen in the photonics industry.

Semiconductor lasers intrinsically offer the merits of high efficiency, high simplicity, and high-volume cost-efficiency, making them the top choice for automotive lidar develop­ers. And the market share for these devices will continue to expand as lidar is further commercialized — especially as ongoing advancements continue to address current shortfalls in the technology.

Meet the author

Leon Li is deputy general manager of the Automotive Business Unit at Focuslight Technologies Inc. He previously served as team leader of international sales and later as head of Focuslight’s Open Package Diode Lasers product line before he led a dedicated team to develop technologies and products for the lidar market; email: .

https://www.photonics.com/Articles/Advancements_in_Diode_Lasers_Fuel_Automotive_Lidar/a67732

[Dietl]

Cowan snapshot

 

 

[Dietl]

Ouster Automotive slide