Month: January 2023

Heading into 2023, LiDAR (Light Detection and Ranging – a 3D imaging technology) appeared to be struggling, given the delays in launching profitable autonomous vehicles and the ~80-90{515baef3fee8ea94d67a98a2b336e0215adf67d225b0e21a4f5c9b13e8fbd502} drop in valuations of SPAC-funded (Special Purpose Acquisition Company) LiDAR companies. LiDAR pioneers were facing severe business hurdles. Some declared bankruptcy (Quanergy, Ibeo), shut down operations (Argo.ai’s LiDAR and AV unit) or announced a merger with a competitor (Velodyne with Ouster). More consolidation is undoubtedly on the horizon since the market opportunities cannot support the number of LiDAR companies that sprouted during the 2015-2020 timeframe (> 70).

Las Vegas spawns optimism, however, and is probably one of the reasons CES is held there. The general buzz around autonomy and LiDAR was upbeat and surprisingly realistic.

OEMs, trucking and industrial equipment providers were bullish on limited passenger car autonomy (L2+ and L3) and L4 for trucking and off-road applications. L2+ refers to an enhanced driver-assist capability that includes automatic lane-keeping and lane-change maneuvers. The human driver is still in control (mind on). L3 is an autonomy feature where a car can drive without human attention in specific conditions (geography, speed, weather, etc.). The driver needs to take over within 10 seconds of being cautioned to do so. L4 refers to full autonomy in specific geographies, with no human driver required. L4 ride-hailing for cars with paying customers is ongoing in fair-weather urban locations (San Francisco, Phoenix, Austin). However, the business case for this segment and the ability to expand to adult cities like New York, Chicago and Boston is unclear. L4 for highway trucking and last-mile deliveries are more tangible. L2, L2+, L3 and L4-trucking are good news for LiDAR companies since they require road-facing 3D sensing functionality for perception and localization. Inside-cabin 3D sensing for passenger safety monitoring is also on the rise.

Significant CES 2023-related autonomy announcements:

1. Daimler announced the launch of L3 capable cars in the US (after a 2022 Europe launch). The Drive Pilot functionality will be available on the EQS and S-Class. Valeo, a France-based automotive Tier 1 supplier, is the current LiDAR provider.
2. Volvo is accepting preorders for the 2024 EX90. It incorporates Luminar’s LiDAR as standard equipment to support driver-assist functions. It is unclear whether it will support L2+ or L3 functionality.
3. Mobileye announced the launch of the Mobileye SuperVision™ – an advanced driver assistance system. The system uses sensors (including multiple LiDARs), positioning and mapping/other information from their deployments over millions of vehicles. Mobileye is a pioneer in the autonomy space, with sizeable revenues (> $1B annually) and close to profitability. It expects to launch L4 services in Israel and other locations in 2023.
4. Sony and Honda announced a collaboration to launch the Afeela electric vehicle with L3 capability. Honda has prior experience with L3 autonomy trials for which it used Valeo LiDARs.
5. Oxbotica, based in the United Kingdom, raised $140M to scale its autonomy platform to support autonomous operations for people and goods movement in campus-style environments like industrial complexes, airports, etc. They use multiple LiDARs for perception (Hesai is one of the providers).
6. Trucking autonomy is on the rise. Waabi, Waymo, Aurora, Gatik, Plus and Kodiak Robotics showcased their automated driving systems to support L4 trucking. All use LiDAR.
7. Apple seems to have postponed plans for a fully autonomous car and embraced the idea of offering limited autonomy for highway driving (L3?) in the shorter term. If true (you never know ! they were not at CES), it is good news for LiDAR suppliers, although launching a car will take time.
8. Caterpillar and John-Deere, leaders in the mining/earth-moving and agriculture equipment sectors, are bullish on deploying autonomy platforms (see figure above).
9. Autonomy at sea is becoming a thing. HD Hyundai (HD = Human Dynamic), a $60B/year revenue company engaged in shipbuilding and energy, performed a successful L2 level autonomy trial for a transcontinental voyage of an LNG tanker in 2022. The company had a massive booth at CES focused on areas like green shipping, energy source development and autonomous shipping. Kisun Chung, the CEO, aims to “democratize pleasure boating for the masses” by providing AI-based navigation and docking capabilities. Perception in ocean environments is incredibly challenging because of wave and wind motion and its impact on image temporal stability. The latency between control and positioning is also high, demanding long-range perception with low latency. Current trials deploy a 200 m range, surround view 360⁰ rotational LiDAR. Commercialization will require LiDAR prices to reduce to accelerate affordable deployment.
10. Tesla: offered convenient rides through the Tesla tunnel to move attendees through different exhibit areas of CES. If only Elon Musk would agree to use LiDAR. Oh well! At least they have warmed to using radar again. LiDAR next???

Brian Collie is the Global Automotive Lead at the Boston Consulting Group. He is bullish on driver assist (L2 and L2+ in passenger cars), trucking autonomy (L4) and, in the longer term, L4-level autonomous ride-hailing. According to Mr. Collie:

• The viability of L3 autonomy is not as clear for two reasons. First, it is questionable whether customers will pay a premium for the incremental value created by L3 (vs. L2 or L2+). Second, there is a high risk of unintended consequences as disengaged drivers become complacent, leading to safety incidents.
• The timeline for L4 for ride-hailing is taking longer than many predicted a few years ago. Commercial scaling is not expected to occur until the next decade as the technology matures and gains customer acceptance. Costs for an L4 vehicle (with a complete autonomy stack) need to be < $80K for a viable business case.
• Trucking autonomy (L4) is compelling now. The economics are attractive, and it solves driver shortage and other issues impacting goods movement today.

Richard Bishop, a pioneer and expert on trucking autonomy, presented an excellent analysis of why he believes transportation autonomy is very much alive. The three-part article, the last of which was released just before CES 2023, provides an objective analysis of use cases that can leverage autonomy – in off-public-road areas (mining, logistics yards), highway trucking and last-mile delivery.

The excitement, investments, trials and product launches in the autonomy space directly translate to the LiDAR landscape. The 3D functionality that LiDAR brings offers significantly superior perception and localization capabilities. Advances in this area are crucial for enabling movement autonomy initiatives. Participation by US and international LiDAR companies was strong at CES 2023. New product features, manufacturing scalability, price, customer wins, miniaturization, and in-cabin 3D sensing were critical discussion areas.

Further consolidation/bankruptcies in this space are inevitable, but this should be seen more as a sign of maturity and sustainability of the survivors. It’s not a roulette wheel anymore – companies who will make it through are becoming more apparent.

Automotive services and Tier 1 suppliers displayed new LiDAR products for road-facing and inside-cabin applications. Established LiDAR players with robust products and the ability to scale high-volume production declared design wins with automotive companies and highlighted their manufacturing maturity. New players with innovative approaches to chip-scale LiDAR are emerging. LiDAR applications beyond automotive are becoming a thing – in mining and agriculture. LiDAR is, well, still exciting.

Established & Large Companies Launch New LiDAR Products

In a sign that LiDAR is critical for autonomy, a few large established companies announced that they are entering the LiDAR space. They presumably see gaps they can exploit (relationships, scalability, durability and, importantly, internally available funding resources). It will take time for their products to be field-deployed. Their entrance into a crowded business landscape is a good signal for the health of the LiDAR market. Entrenched LiDAR players should be worried – execution delays will be expensive, and cost and profitability will be under pressure.

Major announcements of LiDAR market entry include:

1. Bosch: is a major automotive Tier 1 supplier and an industrial products powerhouse. Although rumored to be internally developing a LiDAR and investing in Hesai last year, CES 2023 seems to be the first time they displayed real-time operation of their internally developed LiDAR. According to Kay Stepper, Senior Vice President of Autonomy and Driver Assistance, the 250 m range LiDAR supports ~4M points/second resolution, operates at a 905 nm wavelength and uses a rotating mirror to scan the 120⁰ x 18⁰ Field of View (FoV). It supports L3 autonomy for their OEM customers. The LiDAR size is ~40 mm (H) X 150 mm (W) X 100 mm (D).
2. Mobileye: by far the most successful company in the autonomy space today (in terms of revenues, deployments and profitability), the company transitioned from being an internal division of Intel to a stand-alone public company in Q42022. Traditionally camera-centric in its offerings of autonomy products (hardware and software) to OEMs and direct-to-consumer sales for after-market integration, Mobileye has acknowledged over the past two years the need for LiDAR as the level of autonomy in a car increases. Apart from the launch of the Mobileye SuperVision™ product (see previous section), the company demonstrated its internally developed LiDAR at CES. Operating at 1320 nm wavelength (the so-called O Band in which Intel has significant semiconductor manufacturing experience for communication applications), it is a coherent FMCW LiDAR that enables simultaneous range and velocity measurements. Higher wavelengths offer increased eye safety margins and better range performance. The horizontal FoV is addressed with optomechanical scanning. Transmit and receive optics leverage Intel’s chip-scale silicon photonics expertise. The prototype unit measures ~50 mm (H) X 150 mm (W) X 100 mm (D), which is surprisingly bulky considering the use of chip-scale silicon photonics (although this may shrink as productization progresses over the next two years).
3. Sony: announced a joint car development project with Honda that offers L3 capabilities (see previous section). Separately, the semiconductor division demonstrated a variety of road-facing and in-cabin LiDAR products for automotive deployment. Road-facing LiDAR prototypes use edge-emitting lasers and SPAD (Single Photon Avalanche Photodiode) arrays operating at 905 nm. A rotating polynomial scans the FoV and provides a 150 m range and ~1.2M points/second resolution. Sony’s business focus is to sell the semiconductor SPAD arrays to LiDAR companies, a significant development since it brings to bear the massive scale of the consumer smartphone LiDAR market (that Sony currently dominates) to automotive deployment. Sony has also supplied lower resolution, short-range LED-based iToF (indirect Time-of-Flight) LiDAR units to BMW for inside cabin applications. At CES, the company demonstrated an expanded line of such sensors, including units mounted on exterior entry door panels for face ID-verified vehicle entry and inside the cabin to monitor driver attention, passenger safety and seat-belt compliance. iToF units operate at 850 or 940 nm, use SPAD and VCSEL (Vertical Cavity Surface Emitting Lasers) arrays, and are available in VGA formats (640×480 pixels).

Smaller LiDAR Companies are Innovating

As larger companies enter the market and entrenched players worry about productization, stock market expectations and managing design-in wins, smaller LiDAR companies are innovating and offering compelling alternatives in terms of architecture, operating principles, cost and size. Eliminating moving parts in the LiDAR seems to be a priority, accompanied by a significant degree of miniaturization, which is critical for integration and styling in automotive applications. Progress in these directions is also essential for other reasons – as LiDARs become smaller and more chip-like, the number of piece parts in the Bill of Materials (BOM) reduces, the level of functional integration increases and semiconductor-level manufacturing efficiencies kick in. Many precise alignments and calibrations between different discrete components are eliminated. All of this leads to reduced costs, higher quality and production scalability. Performance may lag initially but will eventually be on par with the best LiDARs. Riding the performance curve upward is typically easier than squeezing the cost and size curve downwards (think Clayton Christensen and disruptive innovation).

Examples of these companies at CES include:

1. PreAct Technologies: led by CEO Paul Drysch, the company announced Series B funding ($14M) and the acquisition of Spain-based Gestoos (an AI-based computer vision and gesture recognition company). Spun out of Artis in 2018, (a developer of sensor fusion and AI-based Active Protection Systems like the Iron Curtain for the US Department of Defense), PreAct is developing short-range, iToF-based LiDAR for applications like in-cabin sensing, drone navigation, self-parking and pre-crash detection. LiDAR functionality is based on a CCD array and LED illumination source operating at 950 nm wavelength. The TrueSense™ T30P combines a flash LiDAR coaligned with an 8MP RGB camera for robust edge processing applications. Designed for road-facing, short-range (<50 m) applications, it covers a large FoV (108° x 77°). The frame rate is probably the highest in the business at 125 Hz (critical for short-range perception and control), delivering an impressive 10M points/second (pps) resolution. It achieves all of this in a 45 mm (H) x 30 mm (D) x 110 mm (W) package with an IP-65 rating. The Mojave™ is a LiDAR-only product that addresses a lower-cost, low-range, size-sensitive markets like inside-cabin, drone and factory automation applications (< $100, 25 mm (H) x 25 mm (D) x 90 mm (W)). It uses the same CCD and LED components as the T30P, achieves a 5 m range and addresses a 72° x 54° FoV. Mr. Drysch is excited about “the PreAct team. While many LiDAR companies are struggling, we are growing exponentially, which is appreciated by recent and prior investors and our growing customer base.”
2. Analog Photonics: based in Boston, MA, was founded in 2014 by ex-MIT professor Michael Watts. It has secured funding contracts from DARPA and commercial entities (Tier 1 automotive suppliers) to develop a solid-state, chip scale, 1550 nm coherent LiDAR. This enables the company to operate on a cash-flow-positive basis without raising venture money. Scanning is achieved by wavelength dispersion in the VFoV and a dense array of optical antennas to address the HFoV. A unique aspect of this LiDAR is that no external focusing optics are required for transmit/receive directions since the scanning function is achieved on-chip with well-collimated beams. The prototype displayed at CES was a short-range, chip-scale LiDAR manufactured using silicon photonics-based integration of the transmit, receive and scanning functions. It has a range of 30 m, addresses a 90° x 80° FoV and delivers a 2M pps resolution. The footprint is impressively compact – 25 mm x 25 mm x 10 mm (probably the smallest volume LiDAR currently for the performance delivered). A-sample products are planned for release in 2025, and automotive series production versions in 2027. A long-range version is under development, targeting performance 250 m range, 120° x 30° FoV and 15 M pps resolution.
3. Trieye: based in Israel, has developed a 3D SWIR imaging and depth sensing solution they call SEDAR (Spectrum Enhanced Detection and Ranging). Founded in 2017 and currently staffed by over 100 people, the company creates CMOS-based SWIR (Short Wave Infrared) imaging arrays fabricated using a high-volume manufacturing platform. Unlike the expensive InP processing typically used for SWIR imaging, this approach provides a breakthrough in cost (Trieye claims 90{515baef3fee8ea94d67a98a2b336e0215adf67d225b0e21a4f5c9b13e8fbd502} cost reduction), resolution and depth sensing capabilities. In concert with shutter control electronics, a high-power, solid-state pumped laser in the 13XX-14XX nm wavelength range enables gated imaging from which 3D information can be extracted. The advantages are eye-safe flash operation with no moving parts, a high-resolution LiDAR depth map fused with a SWIR image (both with 1280×1024 pixel resolution), a range greater than 200 m at 30⁰ x 15 ⁰ FoV, and a compact footprint. Prototypes exhibited at CES measure 70 mm x 30 mm x 50 mm. OEM and Tier 1 customers would incorporate the individual components (laser, detector) to optimize their integration and location needs. The DSP and depth sensing algorithms can be incorporated into their existing ECU.
4. Ouster: is not a “small” company – it is publicly listed and has significant revenues ($45-55M/year), a majority of it in the non-automotive space. It is a consolidator in the LiDAR space (they acquired flash LiDAR maker Sense Photonics in late 2021 and are expected to consummate a merger with arch-rival Velodyne sometime in 2023). They are listed here because the VCSEL-SPAD flash LiDAR product based on Sense’s technology is their recent entry into the automotive market. Internal development has accelerated since the acquisition, and the company launched three versions of the flash LiDAR product at CES, specifically to address the automotive market. According to Angus Pacala, the CEO: “A common remark I hear is that flash lidar is more affordable, more compact, more durable with no moving parts and more scalable, but is decades away and cannot currently meet the performance needed. Well, our digital flash sensor suite is here. It is a game changer and an essential part of Ouster’s strategy to capture the automotive market”. The long-range solid-state unit has a range of over 200m at 10{515baef3fee8ea94d67a98a2b336e0215adf67d225b0e21a4f5c9b13e8fbd502}. It is designed to be mounted behind the windshield, roof or behind-the-grille and measures a compact 130 mm (W) x 40 mm (H) x 80 mm (D). The short and mid-range sensors also share a height of 40 mm, with a lower width dimension. A-prototypes are currently available and undergoing testing by automotive customers. Series programs are slated for production in 2025/2026.

Mature LiDAR Companies are Focused on Automotive Scale Manufacturing and Quality

A few LiDAR companies have secured well-publicized design wins with major OEMs. Their immediate focus is on successful platform integration, manufacturing scale-up and achieving automotive-grade durability and quality. They are also developing new products/platforms that are modest extensions of their current products. With the emergence of new well-funded, and prominent players entering the LiDAR space, they must execute flawlessly on their design wins. But they also need to watch the more recent, nimbler and smaller players to ensure they do not get disrupted by the emergence of solid-state LiDAR with no moving parts, with price, size and reliability/durability benefits.

1. Valeo: is designed into the Mercedes S-Class with its SCALA 2 LiDAR (currently in production) to support its L3 autonomy feature. Its next-generation version, the SCALA 3 uses a similar platform as the SCALA 2, but delivers significantly higher range and resolution performance. (SCALA 2 addresses 80 m range and 300,000 pps in a ~ 600 cm³ volume with a square format. SCALA 3 doubles the range and delivers 10M pps in a ~1000 cm³ volume, with a 45 mm height and rectangular format suitable for roof or behind-the-grill mounting). The higher performance of the SCALA 3 is expected to double the speeds at which the L3 autonomy feature can perform. Valeo is also attempting to standardize the SCALA 3 and minimize/eliminate customization to create economies of scale. SCALA 3 is expected to start production by early 2025 and has been selected by Stellantis and an Asian OEM to support L3 autonomy features.
2. Innoviz: has been successful in production wins with BMW for its InnovizOne LiDAR and with Cariad (the software arm of Volkswagen) for InnovizTwo, which has completed its development and is in production. Innoviz One delivers 120 m range and ~3M pps resolution in a ~500 cm³ volume and is in production with Magna, a Tier 1 supplier. Innoviz Two delivers 200 m range and ~10 M pps resolution in a similar volume. Low-to-mid volume production is done at it’s Israel location in concert with Jabil (Europe-based, automotive-certified contract manufacturer). The company also introduced a 360⁰ surround view LiDAR (Innoviz360) at CES to address the trucking and fleet autonomy markets. This product will be available for sale in Q2 2023.
3. Hesai: announced a design/production win with SAIC’s Rising Auto brand at CES for its long-range LiDAR, the AT128. Designed for driver assistance functions (L2 and L2+) and L3/L4 autonomy levels, Hesai claims it has won orders of several million units of expected volume from 10 OEMs (primarily based in China) and L4 autonomy providers like Zoox and Nuro. The AT128 started volume production in 2022 and is currently shipping at over 20,000 units/month, with 60,000 units delivered. It provides a 200 m range and a resolution of 1.5M pps. Hesai refers to it as a “hybrid” solid-state LiDAR (probably uses an optomechanical scan across the HFoV). Hesai also launched the FT120, a fully solid-state lidar (it’s unclear how) for near-range blind spot coverage. It addresses a 100° x 75° ultra-wide field of view (FOV), 100 m range and a resolution of ~200,000 pps. The AT128 measures 137 mm x 112 mm x 48 mm. The FT120 is much more compact at 68 mm x 75 mm x 90 mm. Hesai is expanding its manufacturing capabilities to address volumes of 1M units/year.
4. Luminar has announced past production design wins with Volvo and Polestar for their 1550 nm, pulsed ToF LiDAR. As mentioned above, Volvo announced that its 2024 EX90 would incorporate Luminar’s LiDAR as standard equipment to support driver-assist functions. Luminar also announced a design win with SAIC’s Rising Auto brand (Hesai announced a similar win with the same customer). Both suppliers (Luminar and Hesai) may have been selected for different autonomy applications. Volvo and SAIC vehicles with Luminar’s LiDAR were displayed prominently at their CES booth.
5. Innovusion: is similar to Luminar in terms of technology and product capabilities. Their primary play seems to be in the China OEM market (Nio and others).
6. Cepton, already selected as a LiDAR provider for General Motors 2023 Ultra Cruise models with L2+ functionality, closed $100M in funding from its manufacturing partner, Koito.
7. Aeva, Robosense and Benewake had a prominent presence at CES.

Las Vegas is an idea CES venue. The city inspires feelings of hope, risks and payoffs. It shows what is possible and is an optimistic beginning to every New Year. I’m betting it will hold for LiDAR in 2023 and beyond.