LiDAR Goes Back To The Future

　　LiDAR is emerging as an increasingly important piece of the enabling technology in autonomous driving, along with advanced computer vision and radar sensor chips. But LiDAR systems also are finding their way into a variety of other applications, including industrial automation, including robotics, and unmanned aerial vehicles.

　　Advanced mapping is another rapidly growing market for LiDAR, which is not entirely surprising considering this was the original application for this technology. But work is progressing in this area as mapping grows more powerful and sophisticated due to new developments in LiDAR technology. In fact, LiDAR was used to locate a vast Mayan metropolis in the jungles of Guatemala, which included thousands of structures and extensive causeways for commerce that were hidden from view for centuries.

　　These expanding use cases, along with continued form factor improvements in the technology, bode well for the LiDAR market. ABI Research forecasts the market for automotive LiDAR alone will be worth almost $13 billion by 2027.

　　And with LiDAR growth, comes growth in related markets. Yole Développement predicts the market for gallium nitride power devices, for example, will enjoy a compound annual growth rate of 79% over the next five years, reaching $460 million by the end of 2022. Power GaN technology is well-suited for high-performance and high-frequency uses, the market research firm says. “LiDAR, wireless power, and envelope tracking are high-end low/medium voltage applications, and GaN is the only existing technology able to meet their requirements,” notes Yole’s Ana Villamor.

　　So while leading LiDAR vendors are pursuing their vision of autonomous vehicles, they are shipping products today for drones, industrial systems, and mapping, in addition to developing advanced driver-assistance systems.

　　Autonomous and assisted driving

　　Still, the biggest opportunity for LiDAR is automotive, and work is underway to reduce the number of moving parts and the cost of these systems with solid-state designs.

　　“To me, LiDAR is a really interesting system because there are so interesting components in there that all have relevance to semiconductors,” says Jeff Miller, product strategist for Mentor, a Siemens Business. “There’s some very interesting work being done in the silicon photonics area to try to make solid-state LiDARs. And there’s some very interesting work in the more traditional LiDAR space in terms of how do I get power to the laser, how do I control my laser pulses, and very interesting power transistor designs. And then there’s the data processing angle on this. What do I do with this enormous volume of data I’m producing in these 3D point clouds that come off the LiDAR sensor? How do I make any kind of sense of that? Ultimately, the value of this is providing information about what my car can drive through and what my car can’t drive through. That’s ultimately what we’re after with these sensors, because the primary end-vision product for these things is the driverless car.”

　　Early automotive LiDAR systems were spinning domes on the roofs of prototype cars. That won’t fly in vehicles for the consumer market, of course. “Commercialization is already starting to happen,” Miller observes. “It needs to look like a regular car.”

　　And that’s where things really start to get interesting for LiDAR. “The race is on for a low-cost, small-form-factor, solid-state LiDAR,”says Ian Dennison, senior group director of research and development for the Custom and IC Packaging Group at Cadence, who notes that the price tag needs to drop to $200 or less. “There is a bit of a gold rush going on. There are plenty of startups, plenty of investment. Part of the excitement is that there’s quite a variety of silicon design fabrics in the mix. You’ve got CMOS, for sure, but you’ve also got silicon photonics, you’ve got silicon MEMS. Different players have different perspectives on which part the technology mix should be.”

　　Dennison adds that LiDAR offers excellent depth resolution and “a great benefit for the characterization of an autonomous vehicle. It’s the real killer feature for LiDAR.”

　　What’s different

　　The growth in LiDAR also is boosting demand for high-current, high-frequency, low-resistance, low-capacitance power transistors, made with GaN and other exotic semiconductor materials.

　　“We are seeing a trend toward more integration to bring higher performance attained in III/V materials into traditionally silicon systems,” says Chris Cone, a product marketing manager at Mentor. “LiDAR design is highly sensitive to geometry and requires many of the same capabilities developed for silicon photonic IC design, including the ability to generate design components based on curvilinear objects, the ability to assemble design components with multiple waveguide types along with the required tapers and transitions, and multi-domain circuit simulation with proper modeling of optical, electrical behavior. Very importantly, physical verification needs to be photonic structure-aware, as standard verification is prone to produce large amounts of false errors. Using standard DRC checking techniques will generate large amounts of false positives that can’t be adequately reviewed. They have no choice to waive large sets of checks, which allows real errors to make it to mask.”

　　As a result, tools vendors are looking at different methodologies. “LiDAR requires new design considerations, which will requires design tools to be flexible to account for new methodologies,” Cone says. “But you can apply the same well-known development milestones to LiDAR design – design completion, design analysis through simulation and layout verification, including DRC, LVS and post-layout extraction.”

　　Some experts don’t characterize MEMS devices as solid-state components because they have moving parts, and some LiDAR vendors have turned to phased arrays instead of MEMS.

　　“Different people take different approaches,” says Cadence’s Dennison. “Silicon photonics is an excellent fit for a problem like LiDAR, a light-based application like LiDAR. If you can use the photonics to manipulate the light, you’re already getting onto a solution that isn’t trying to keep up with the frequencies of light. It’s using light natively.”

　　Mixing MEMS with photonics and CMOS chips presents a co-design problem, he notes. “It’s blended technologies.”

　　No shortage of investors

　　As prospects for LiDAR increase, so does funding for this technology. Three of the leading vendors of LiDAR systems are all based in Silicon Valley – Velodyne LiDAR, Quanergy Systems, and Cepton Technologies. All have been expanding their operations in the past year.

　　While these companies are maturing and readying LiDAR systems for the big automotive manufacturers and their Tier 1 suppliers, they are being pursued by well-funded startups. Aurora Innovation just picked up $90 million in Series A funding, co-led by Greylock Partners and Index Ventures. May Mobility of Ann Arbor, Mich., received $11.5 million in seed funding from BMW i Ventures and Toyota AI Ventures. Seattle-based SEEVA Technologies got $2 million from Revolution’s Rise of the Rest Seed Fund and other investors. Nuro, which is developing a self-driving delivery vehicle, recently disclosed receiving $92 million in a Series A round from Banyan Capital and Greylock Partners. And LeddarTech of Quebec City, Quebec, Canada, last year raised $101 million in Series C funding led by Osram, joined by Delphi Automotive, Magneti Marelli, and Integrated Device Technology.

　　“We see the market as asking for two things,” says Anand Gopalan, CTO of San Jose, Calif.-based Velodyne, high-performance sensing for robotaxis and a smaller-form-factor sensor for ADAS applications. But he also sees industrial applications as a fast-growing market for LiDAR systems, along with drones and mapping.

　　“The mapping market continues to evolve and grow,” he says. “Our focus is shifting from simply providing high-definition maps to providing high-definition maps with a view of enabling autonomous driving. We continue to see that the intermapping market, having a 360° field of view combined with a really accurate LiDAR is pretty important, obviously, to take a good high-definition map. We continue to see a lot of growth in that space. People are finding a lot of interesting uses for LiDAR across the board, like area mapping. It’s exciting for LiDAR as a domain, in general, to see all the new applications. Definitely, in the ADAS space, LiDAR has become accepted as a piece of modality for most ADAS systems. Electric vehicle and ADAS programs around the world all are using LiDAR.”

　　LiDAR is particularly useful when there is snow on the roadway, covering up the lane markings, making it difficult for computer vision to work.

　　One of the big challenges in developing automotive-grade LiDAR are the environmental issues, such as operating in temperatures of 105° C, and up to 150° C, along with conditions of -40° C, according to the Velodyne CTO. That will be critical for automotive applications, which in turn will drive other applications and capabilities.

　　“As you start seeing this first wave of LiDAR being deployed, you will also start seeing more and more smarts or intelligence being baked into these sensors – providing high levels of functionality, like localization or object tracking. The amount of compute in the LiDAR will also constantly increase. That’s the other domain where you will see more work,

　　more interesting products come out in the next couple of years, in the domain of embedding intelligence within the sensor,” Gopalan says.

　　Quanergy CEO Louay Eldada also is seeing a bump up in the LiDAR market. “Things are becoming real. It’s becoming easier to cut through the noise,” he says. “We are getting some large contracts in automotive security, industrial automation, as well as mapping, terrestrial and aerial mapping from drones.”

　　Quanergy builds solid-state LiDAR products, employing optical phase arrays. The goal is to get the price of a LiDAR system-on-a-chip device down to $100, according to Eldada. “LiDAR is the primary sensor you must have for Level 4. We want to support the automotive industry.”

　　There are an estimated 50 companies in the LiDAR business, and that field will be culled in the years to come. But in the short term, new companies are still cropping up. Cepton Technologies was established in 2016. Mark McCord, the startup’s vice president of engineering, says his team is working on longer-range sensors for automotive applications, and the company expects to turn out automotive-grade products in the second half of this year.

　　Wei Wei, Cepton’s director of business development, says of the 50 companies in the LiDAR market, less than five can deliver LiDAR systems with a range of 200 or more meters. He sees more carmakers coming out of the prototype stage with their most advanced vehicles by the end of this year.

　　Conclusion

　　The LiDAR market is still in a nascent stage, as far as automotive electronics are concerned. But the market for this technology is growing well beyond a single vertical market. Mapping has returned as an important application for LiDAR technology. The sensor’s use in drones and robots could be lucrative for some companies, as well.

　　Considering this technology was relatively obscure several years ago, that’s a big change. And as the price continues to drop, it will show up in even more applications. For LiDAR, the real growth phase is still to come.