The Big Bend Theory And Beyond- Are We There Yet?

In a recent article, I discussed the factors driving the consolidation in the automotive LiDAR industry. I also presented 4 areas ripe for investments, one of which was solid-state LiDAR. The focus of this article is on assessing the approaches and maturity of LiDAR that uses solid-state scanning techniques for beam steering and enables efficient use of photons in space and time.

For the purpose of this article, a solid-state LiDAR is one that uses non-mechanical methods to project and collect photons across the Field of View (FoV). MEMs beam steering and other mechanical scanning techniques are excluded. 2D Flash LiDAR qualifies as a solid-state LiDAR. However, it is is fairly mature (Continental, Leddartech) and limited in performance, and is not included in the discussion here. 

The advantages of solid-state LiDAR include reliability, immunity from shock and vibration, scalability, and lower manufacturing costs. Features like Region of Interest (ROI) scanning and higher resolution in the ROI can be realized. Stylistic blending of the LiDAR into the car body design becomes possible which is important for the eventual purchase of L4 and L5 cars by consumers rather than transportation providers (this will take a while but is important to OEMs lusting after the 100M cars/year market for purchases by consumers).

The term “solid-state LiDAR” has been overhyped and over-promised, with not much to show in actual practice until recently. The frenzy that surrounded this space starting 4 years ago contributed to this, as did the fact that it is a difficult problem involving fundamental material science innovations and expensive process optimizations which need significant funding. A small portion of the >$2B in LiDAR funding to date has gone towards serious innovation in this area. Hopefully, that changes as the technology gathers momentum and matures along the learning curve, and expectations become more realistic.

I concluded that the best way to present this topic was to interview key leaders in this area, and motivate them to discuss their approaches in terms of where they are, where they need to go and when they will get there.  I interviewed 4 companies that approach solid-state LiDAR in different ways – technically and from a market perspective. They include Quanergy, Lumotive, Draper Labs and Baraja. I would like to thank them for being transparent and sharing the level of detail they have. 

QUANERGY

Based in Silicon Valley, Quanergy essentially coined the phrase “Solid-State LiDAR”. In parallel with its mechanically spinning product, the company continued to pursue solid-state LiDAR based on Optical Phase Arrays (OPA). The system relies on a 2D array of SPADs (Single Photon Avalanche Photodetectors) and 9XX nm single mode laser sources feeding the OPA based emitter which steers the laser beam across the HFOV. The current product generation uses multiple sets of emitters to provide VFOV coverage with Region of Interest (ROI) capability in the horizontal direction. Development is underway on using OPAs for 2D electronic beam steering. A schematic of the beam steering concept is shown below:

Although Quanergy’s early push on solid-state LiDAR was in automotive, recent productization efforts have focused on security and smart space applications that have lower performance thresholds than required in the automotive industry. The company had announced in 2016 its intentions to deploy such solutions in drone, security, robotics, industrial automation and people counting applications. More recently, Quanergy highlighted an access control security product which uses a solid-state LiDAR in conjunction with it’s QORTEX People Counting and Access Control software for building and campus security applications. The system is installed above regular doorways and Quanergy claims it can achieve 98%+ people counting accuracy under all lighting conditions.

I asked Tianyue Yu, Quanergy’s CTO and co-founder to comment on the earlier claims of a $250 LiDAR and whether this was achievable and whether the manufacturing and implementation challenges of being able to fabricate low-cost OPAs which enable low-cost diode lasers to be used are now feasible. ”In 2019 we commercially launched the industries’ first CMOS-based solid-state LiDAR product line featuring an intelligent Qortex solution embedded in the S3-2 sensor. Our next target application is for the industrial automation market, which requires long ranges, wider FOV, and a variety of configurations. We believe that, as we broaden the market coverage and increase product volume, we will gradually perfect the fabrication of OPA technology and enable volume manufacturing of low-cost, solid-state LiDARs. As the technology matures and performance improves, we will address the automotive industry with cost competitive reliable products.

LUMOTIVE

Based in Seattle, Washington, the company counts Bill Gates as one of their investors. Lumotive’s solution is based on an all CMOS construction, except for the laser diode chip.  The architecture illuminates the entire VFOV, which is then scanned across the HFOV using LCM (Liquid Crystal Metasurfaces), a technology similar to imaging radar (Metawave, Echodyne). LCMs operate by crowding thousands of tunable optical resonators onto the surface of a CMOS semiconductor chip. By tuning the characteristics of each resonator using liquid crystals, a phase profile can be created across the surface of the chip which causes incident photons to be reflected in the desired direction. In this way, the LCM acts as a programmable mirror but controlled electronically with no moving parts. A schematic is shown below:

Lumotive’s current ToF system operates at 905 nm and can use edge emitting lasers (EELs) or Vertical Cavity Surface Emitting Lasers (VCSELS). The laser itself according to Lumotive is a low-cost source (could be single or multi-emitter). On the detector front, the LCMs de-steer reflected photons from the FoV onto a 1D SPAD detector array. Region of interest adaptive scanning is also supported. 

The LCM approach can also be used with a 2D detector array operated in a rolling shutter configuration, operate at 15XX nm wavelengths and in FMCW LiDAR systems.  The SPAD detector, LTM scanning, and laser diode driver are all manufactured in CMOS for the 9XX nm system, enabling low-cost manufacturing and volume scale-up. A recent press release by Lumotive advertises their product launch plans:

 “With samples available in the fourth quarter of this year, the Lumotive X20™ and Lumotive Z20™ LiDAR systems target the automotive and industrial automation markets, respectively. Lumotive’s M20™, addressing the needs of the consumer and mobile markets, will be introduced in 2021. The X20 targets long-range automotive applications with range over 120 meters in bright sunlight and a 120° x 30° field of view. The Z20 will have a shorter range (~ 50 meters) but an expanded 70° vertical field of view to address industrial and short-range automotive needs.”

I asked Lumotive CEO, Bill Colleran to comment on pricing, design-in’s and customer traction to date for different applications. “The huge challenge of LiDAR is balancing performance, characterized by range, resolution, and frame rate, with commercial viability as measured by size, reliability, and cost. We’ve been able to achieve new levels in all these areas without significant compromise in any. One size doesn’t fit all when deploying 3D sensing in different applications, and our underlying architecture allows us to efficiently scale the technology to meet specific market requirements while maintaining the cost benefits we achieve through our manufacturing approach.”

DRAPER

Yes, this is the legendary Charles Stark Draper Laboratory, based smack in Red Sox territory in Cambridge, MA.  And while I am a passionate Yankees fan, and generally hate anything Red Sox, I felt compelled to include them in this discussion.

Draper has been working on an intriguing solid-state beam steering technique, which they claim has applications in automotive, robotics, industrial automation, and even in free space telecommunications (I knew eventually innovations in LiDAR would loop back into the telecom space!) Their technique uses MEMS switches, rather than the more typical MEMS mirrors to steer the beam. The digital design of these switches enables fast operation and allows for a random-access capability and selective FOV illumination. By my definition, MEMs based beam steering is not considered solid-state, but since the Draper solution uses MEMS as a simple digital switch, I feel it qualifies as a solid-state LiDAR.  Steven Spector, one of the primary investigators recently presented his work at a CLEO conference. A schematic of the approach is shown in the figure below:

The PIC requires a single spatial mode diode laser, which in my view can be limiting for longer range application. The technique can work either at the 8XX-9XX nm or 13XX-15XX nm wavelengths and can support both ToF and FMCW coherent LiDAR. This allows for a flexible tool set for different LiDAR architectures and applications. The laser sources do not have to be coherent for ToF LIDAR but will need to be for FMCW implementations. 

Proof of concept of a ToF system have been demonstrated with this technique using 1550 nm laser sources and achieved range performance of 15-50m with 0.1° (1.5 mrad) angular resolution, FOV of 1° x 1°, at frame rates of 5 Hz. Steven Spector claims that Draper’s simulations show that 200 m ranges at 20 Hz and 60° x 20° FOV are achievable. I asked him why readers should believe this large performance extrapolation of ~20,000X. “The path to longer ranges and higher pixel rates is fairly straightforward. Our average laser power (of a few milliwatts) is about 100-1000x less than we expect to have available in an automotive LiDAR and our receiver aperture area is nearly 100x smaller. Our work, so far, has focused on demonstrating the MEMS technology itself. Scaling rules for LiDAR are well understood, so we have high confidence about how our technology will scale. Now that we have demonstrated the MEMS technology, we can turn our attention to building an entire LiDAR system that will meet automotive requirements.”

Draper is admittedly not a manufacturing powerhouse, but it is incredibly strong on technology. One of the advantages they claim is the ability to fabricate the scanning mechanism using low-cost CMOS like processes. Their business model is essentially to license the scanning technology for different applications and work with system providers to realize performance, manufacturability, and cost goals.

BARAJA

Based Down Under with what I think is the coolest name in all of LiDARDOM (that is definitely a word I have coined !), Baraja is the brainchild of technologists with experience in Dense Wavelength Division Multiplexing (DWDM) which fueled telecom optics 20 years ago. Baraja’s Spectrum-Scan technique exploits the wavelength of light and essentially bends it in different directions using an optical element like a prism (Baraja is Spanish for a deck of cards, and as a verb means to shuffle, a nod at how the system shuffles through different wavelengths of light). It requires the use of a C-band tunable laser (1550 nm +/- 20 nm), which they claim can be easily leveraged from the telecommunications industry at low costs. A recent TechCrunch article discusses their system in the autonomous vehicle context. A high-level schematic is shown below:

The Spectrum-Scan technique allows for a maximum of 2000 laser spots (by tuning /shuffling across 2000 wavelengths in the C Band) across a 30° VFOV). These spots can be switched in the order of nanoseconds, and more importantly, can be adaptively foveated across the fast axis (typically the VFOV). This provides a significant resolution advantage in regions of interest (ROI). For the slow axis (typically the HFOV), Baraja indicates it is free to use a variety of reliable methods, such as polygon mirrors or prism scanners (which may make it semi-solid state). In terms of modulation, they use Random Modulated Continuous Wave (RMCW) which uses pseudo-random pulse encoding. The combination of this modulation and the wavelength tuning provides a significant advantage in terms of interference mitigation and spoofing protection. 

In terms of performance, Baraja claims 200 m range performance across a VFOV of 30° and a HFOV of 120°, at frame rates up to 40 Hz. Their roadmap also includes using the RMCW modulation to detect the Doppler velocity of objects in the environment. It is not clear whether the system incorporates a local oscillator to enable some type of heterodyning operation.

Baraja is actively engaged in multiple markets as explained by Nick Langdale-Smith, VP of Business Development “In addition to the long-term opportunity we see in automotive, our customers in the mining, smart infrastructure & resource management sectors value Baraja for it’s high performance without having to sacrifice durability or uptime. The mining industry in particular produces some of the most challenging environments for LiDAR – unpaved roads, extreme shock and vibration, high temperatures, dust and humidity, and low tolerance for equipment downtime

Baraja CEO Federico Collarte comments on the maturity of their system, the traction with automotive customers and other markets, and the targeted price points of their system. “The long-term winners in this space will ultimately be graded on a performance-per-dollar basis. Many of today’s LiDAR cling to outdated scanning techniques that simply cannot attain the chip-level scale crucial to meet automotive performance and cost expectations. Meeting price points in the hundreds of dollars will require investment in integration and new approaches to scanning. Baraja’s patented Spectrum-Scan family of technologies are unique in the marketplace: fresh terrain from which we’re deriving a wealth of innovation, extracting a ton more mileage than legacy areas that are almost exhausted. What better way to approach the scaling challenge than through the integration of the scanning mechanism – wavelength – into the laser itself.”

SO, ARE WE THERE YET??

I think we are. Apart from flash LiDAR applications (Continental and Leddartech), solid-state scanning LiDAR has progressed. Quanergy, I believe, is one of the first to launch such a product in the non-automotive space. Lumotive and Baraja are close behind. And hopefully, Draper’s solution will be productized soon.

It is interesting that all companies are looking at applications other than automotive – given timeline risks for L3 and L4 deployment, this is a smart thing to do. The challenge is finding the applications where solid-state LIDAR has an edge – like size, reliability and fast ROI scanning. 

The business strategies seem to be different – Lumotive is addressing the automotive market first and leveraging this to go after industrial and consumer applications. Quanergy and Baraja are addressing the non-automotive markets first and using this as a springboard to automotive opportunities. CMOS seems to be the enabler towards low-cost solid-state LiDAR, although I don’t think a sub-$1000 product exists as yet (it does for some of the opto-mechanically scanned LiDARs). In terms of automotive maturity, non-solid-state LIDAR is ahead (for example, Valeo, Velodyne, Hesai, and Innoviz). 

Dr. Thomas Jellicoe, Autonomous Driving Lead at TTP (The Technology Partners) has this to say about the maturity of some of the solid-state scanning LiDAR approaches.

There is a long history of photonics technologies that promise superior performance by manipulating light on tiny length scales using features such silicon waveguides and liquid crystal layers. They claim low-cost, leveraging decades of investment in semiconductor processing. In reality, however, the fabrication processes required are often non-standard and require costly, time-consuming optimization which is only economical for very large volumes.

There are few examples of products that have successfully scaled, with liquid crystal displays being an exception. Nevertheless, there has been plenty of progress in photonics processing in recent years and if these technologies are starting to mature then the implications are huge for automotive and other industries.

Speak Your Mind

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Get in Touch

350FansLike
100FollowersFollow
281FollowersFollow
150FollowersFollow

Recommend for You

Oh hi there 👋
It’s nice to meet you.

Subscribe and receive our weekly newsletter packed with awesome articles that really matters to you!

We don’t spam! Read our privacy policy for more info.

You might also like

3M Makes A Lot Of Masks But Trump Can’t...

President Donald Trump says 3M will have a "big...

After 6 Months of Uncertainty and No End in...

Several months into the pandemic, Peter Newell's Palo Alto, Calif.-based innovation consulting firm BMNT,...

Advice From A Young Black Entrepreneur

Many entrepreneurs have had to overcome challenges in early life. Few have faced the...

Reliance Jio to roll out 100 million low-cost phones...

BENGALURU: Reliance Industries Ltd's telecom unit is looking to outsource the manufacturing of over...