Search

Unleashing LiDAR’s potential: a conversation with Innovusion

interviewee

yimin-li_innovusion

Yimin Li

CTO and Co-founder

Innovusion

Interviewer

Pierrick_BOULAY-PBO_YINT

Pierrick Boulay

Senior Technology & Market Analyst, Automotive Semiconductors

yimin-li_innovusion

Yimin Li

CTO and Co-founder

Yimin Li is CTO and Co-founder of Innovusion. Yimin Li has a Ph.D. in Quantum Electronics from Peking University and has 20+ years of experience with electronics, optics, and laser technology, including LiDAR sensors. He was Tech Lead at Velodyne Lidar and managed technology projects at Baidu, Agilent, AOSense, and GE Health.

Pierrick_BOULAY-PBO_YINT

Pierrick Boulay

Senior Technology & Market Analyst, Automotive Semiconductors

He works in the fields of solid-state lighting and lighting systems, carrying out technical, economic, and marketing analyses. In addition, he leads the automotive activities within the company. Pierrick has authored several reports and custom analyses on topics such as automotive lighting, LiDAR, sensing for ADAS vehicles, and VCSELs. Prior to Yole Group, Pierrick has worked in several companies where he developed his knowledge of lighting and automotive. In the past, he has primarily worked in R&D departments on LED lighting applications. Pierrick holds a master of science in electronics at ESEO (Angers, France).

The market for LiDAR in automotive applications is expected to reach US$3.9 billion in 2028 from US$169 million in 2022, representing a 69% Compound Annual Growth Rate (CAGR).

According to Yole Intelligence’s LiDAR for Automotive 2023 report, Innovusion was the leading player in 2022, having a 28% market share. The LiDAR market in automotive is just starting, and actual LiDAR volume could be tripled in 2023, showing the strong dynamics in this market led by Chinese players, whether OEMs or LiDAR manufacturers.

Innovusion (latest news) is a global LiDAR manufacturer developing hybrid solid-state LiDAR based on 1,550nm components and using a fiber laser as a light source. They have a successful partnership with NIO, a Chinese automotive OEM, and their LiDAR is installed on the roof of these cars.

In this context, Pierrick Boulay, Senior Market and Technology Analyst at Yole Intelligence, part of Yole Group, had the opportunity to share its expertise and vision with Yimin Li, CTO and co-founder of Innovusion.

Discover the details of the conversation below.

Pierrick Boulay (PB): Please introduce your company and your position.

Yimin Li (YL): I am Yimin Li, the CTO and co-founder of Innovusion. I have a background in Quantum Electronics, and prior to founding Innovusion, I worked at companies such as Velodyne, Baidu, Agilent, AOSense, and GE Healthcare, accumulating over 20 years of experience and expertise in electronic, optical, and laser technologies, including LiDAR.

I founded Innovusion with Junwei Bao in Silicon Valley in 2016. It’s hard to believe that just seven years later, we are now a global leader in the LiDAR space and that our ability to design and manufacture 1550nm LiDAR at scale is unparalleled. I think that speaks volumes about not just the quality of our underlying technology but, even more so the quality of our global teams. We have R&D teams in the US, in Silicon Valley, as well as in Suzhou & Shanghai, in China. In addition, we also operate state-of-the-art, highly automated, auto-grade manufacturing bases in Suzhou and Ningbo that are fully operational. These were all developed and built from scratch and are producing and delivering at scale for our global automotive partners, like NIO and Faraday Future. For example, our flagship, automotive-grade LiDAR sensor, Falcon, is included as standard equipment on almost all NIO models (including their ET7, ES7, ET5, EC7, ES6, ET5T, and ES8), as part of their highly acclaimed Aquila autonomous driving system.

In 2022, our LiDAR achieved the No.1 volume and revenue in passenger vehicles globally (even surpassing expectations from Yole!). This year is going even stronger, and we’re on target to surpass last year’s passenger vehicle volume by the end of July 2023, with a forecasted trajectory of continued high-speed growth thereafter.

At the same time, we have signed strategic contracts with many commercial vehicle partners, such as TuSimple, Encon, Zhito, Plus, and DeepWay, to jointly promote the large-scale application of high-performance LiDAR in the commercial logistics field.

In addition, our sensors are also vital to smart transportation initiatives and programs, and we’re also partnering with numerous giants of smart transportation, highway, rail, and industrial automation globally in order to help bring the power of LiDAR to improve the efficiency, safety, and flow of traffic in cities and ports around the world.

PB: Could you briefly introduce Innovusion ‘s LiDAR technology?

YL: Our company focuses on the development of hardware and software solutions for LiDAR. Currently, we have the Falcon series and Robin series LiDAR hardware products, as well as OmniVidi on the software side.

Our flagship LiDAR sensors are the Falcon series, which uses 1550nm laser technology. We’ve produced and shipped over 150,000 units of this sensor and are proud to be a critical part of NIO’s Aquila sensor suite and standard equipment on almost all of their new vehicles. These are the only mass-produced 1550nm LiDAR today, and that is something that we are very, very proud of. It is hard enough to get your LiDAR to work in a lab, but to then mass-produce, deliver to your customers like clockwork, and ultimately help power L2+ ADAS systems on cars all around the world? Those are the kinds of things that Junwei and I dreamt of when we started this company, and sometimes it is hard to believe how far we’ve come. From a technical standpoint, Falcon has a maximum detection range of 500 meters (250 meters @ 10% reflectivity), making it ideal for long-range applications. It also features a hybrid solid-state scanning mechanism that allows for high-precision scanning and dynamic focusing. When accounting for the precision, the production at scale, and our ability to work closely with partners and customize the details as needed, Falcon really stands in a class of its own at the moment. And it keeps getting better.

Falcon – Automotive-grade Ultra-long Range Front-view LiDAR

Next is our Robin platform, which has a 905nm laser light source. It also leverages hybrid solid-state technology and incorporates cutting-edge electronics and optical technology and a highly modular architecture to achieve excellent product performance and adaptability in various laser detection scenarios.

Robin comes in two models – Robin-E & Robin-W:

  • Robin-E is an advanced long-range forward-looking LiDAR that currently achieves the leading detection level amongst forward-looking lasers in the industry. It has a maximum detection range of 250 meters and a standard detection range of 180m@10%. It features a resolution of 0.1° × 0.2° and a field of view of 120° × 24°. The overall design is exquisite and compact, with the ability to achieve a curved surface design on the optical window, seamlessly fitting the vehicle’s structure and appearance. It can easily be integrated into different positions, such as headlights and bumpers. In fact, we’re currently collaborating with a leading automotive glass manufacturer to explore new aesthetic and practical installation ideas for our LiDAR, including the co-development of a rear windshield installation solution featuring Robin-E. 
  • Robin-W is the sibling of Robin-E. It’s a high-performance medium- to short-range wide-angle LiDAR and was designed with side and rear installations in mind. The stats are still fantastic. The standard detection range, for example, is 70 meters at 10%. That is twice the ranging capability of similar products in the market right now. The same thing with the resolution: 0.1° × 0.4°, which is significantly higher than the detection accuracy of similar products. This is important because the high resolution provides more accurate target recognition, ensuring clear visibility and enabling early identification and appropriate reactions. According to our calculations, in typical scenarios such as high-speed lane change, unprotected turns in urban areas, and vehicle parking, the detection distance in the side and rear directions needs to be at least 70 meters to meet safety requirements. Those considerations drive our design and inspire us to keep pushing further.
Robin-E – Image-grade Long Range Front-view LiDAR 

Another thing that really sets us apart is that we don’t just design our sensors to have great specs and look great on paper – we design them to be easy to manufacture and, more importantly, easy for our partners to work with and design around. The power consumption, noise levels, and heat dissipation of the Robin line speak to this perfectly. Robin sensors –both Robin-E and Robin-W – have ultra-low power consumption –less than 10w. To put that in context, this is the first time the power consumption of automotive-grade LiDAR has been reduced to single digits. This is significantly lower than the mainstream products in the current forward-looking LiDAR market, and when paired with the low noise level (below 20 dBA) and integrated heat dissipation, Robin-E is a very friendly sensor to work with from an engineering and design integration perspective.

Robin-W – High-performance Mid-to-short Range Wide-FOV LiDAR

But then it gets better. The cherry on top of all this is our OmniVidi software platform. This is perception middleware that serves as a complete perception solution. It incorporates cutting-edge deep learning frameworks and provides a comprehensive toolchain – including algorithm model components, perception function suites, intelligent data, and quantitative evaluation. By combining various advanced technologies and traditional methods, such as the SightNet model for single LiDAR forward perception, lightweight fusion perception model for multiple LiDARs, spatiotemporal fusion 3D object detection and tracking algorithms, clustering, and Kalman filtering, it effectively reduces computational load. It also integrates both BEV and RV dual-mode detectors to strike a balance between real-time performance and accuracy, enabling efficient detection of surrounding objects and delivering outstanding perception capabilities.

PB: LiDAR has traditionally been seen as a high-cost component, especially those based on fiber lasers. What strategies or innovations is your company implementing to make LiDAR solutions more affordable and accessible to automotive manufacturers?

YL: Let me say that I don’t think the true value of LiDAR has even been realized yet. It is an extremely powerful technology with so many potential applications and possible impacts down the road, only a handful of which are known at the moment.

That said, the price. I’ll start by noting that our first focus is, and always will be, the quality of our products – their precision, their performance, their reliability. That said, we are also always very cost-conscious and are continually looking to optimize our products and reduce costs where it makes sense.

We are actively trying to bring down the costs of producing our LiDAR and have already taken many active steps towards that goal. Building our production capabilities and supply chains from scratch meant we needed to invest a lot early on. But as volume increases and efficiencies of scale become evident, there will be significant room for cost reductions. This requires continuous expansion of the ecosystem and ongoing investments, but it also means that many components of 1550nm LiDAR that are currently expensive are beginning to come down in price as our technology advances and production scales up. With this, you will also see the price gap between 1550nm and 905nm LiDAR start shrinking, with our expectation being that the cost difference between the two will eventually be minimal, around 5% to 10%.

In addition to keeping an eye on current costs, we’ve also created and begun to implement a roadmap for meaningful cost reductions and concrete plans for achieving them. In the meantime, Innovusion is outperforming our relevant competition by a wide margin in terms of current performance and cost-effectiveness, and we’ll continue to focus on delivering valuable, full-lifecycle solutions to customers at prices they can afford.

PB: As LiDAR technology advances, how do you anticipate it will impact the overall design and aesthetics of vehicles? Will LiDAR sensors become more discreet and seamlessly integrated?

YL: Innovusion is dedicated to helping our OEM partners achieve their design visions and providing them with the most flexible and least intrusive options for their designs. While this can mean offering higher-precision sensors that consume less power and generate less heat, it also means delivering products that seamlessly fit into the design visions and demands of our partners. To this end, the products we collaborate on with customers undergo intense customization to meet their specific design requirements. While the OEMs will typically take the lead in the design process, we then work closely with them to make those visions a reality. This often means there is a deep degree of customization that needs to happen with the physical design of LiDAR itself, like the curvature of the optical window or the size of the device.

What we really love about this collaborative process, though, is that it can also drive aesthetic design innovations that we are then able to leverage with other products or partners down the road. The power and spirit of this collaboration is perhaps best seen in the NIO ET7 being awarded a Red Dot design award for the smooth integration of LiDAR and other sensors into their autonomous driving capabilities. As the jury noted, ‘The reduced design of the NIO ET7 and the comprehensive use of smart technology merge into a harmonious overall appearance’.

In addition to working closely with our OEM partners to deliver on their vision, we’re also proactively driving the design possibilities of LiDAR through additional research partnerships with major auto suppliers like Fuyao and Wideye. These allow us to explore different installation positions and methods – like installation behind the windshield, within bumpers and headlights, or within the roofline – to better align with the overall aesthetic design of vehicles.

PB: What advancements or developments in LiDAR technology do you believe are necessary to overcome the limitations or challenges currently faced in automotive applications?

YL: There are a number of key areas that we are actively focused on that we believe will help speed up the mass adoption of LiDAR across all manufacturers. First, there is the continuous progress and integration of laser detectors, which will further improve the performance and reduce the cost of LiDAR. In addition, the continuous improvement of LiDAR signal processing algorithms will greatly assist in vehicle control. Finally, as the capabilities of LiDAR continue to develop, there is a constant expansion of the definition of LiDAR usage scenarios and requirements. Currently, front-view lidar products can’t fully meet the demand. In the future, there will be a need for targeted products for new requirements and new scenarios, such as side-view LiDAR.

PB: LiDAR technology has traditionally relied on mechanical scanning systems, but solid-state LiDAR solutions are gaining attention. What advantages and challenges do you see in adopting solid-state LiDAR for automotive applications? How is Innovusion positioned regarding this transition towards solid-state LiDAR?

YL: There has been a lot of buzz around solid-state LiDAR for years, and with good reason. Solid-state LiDAR technology holds the potential for high reliability, but the current technology is still immature and has challenges to solve regarding detection range and field of view. Once these are solved, manufacturing becomes the next challenge and will need to be proven at scale.

In contrast, Falcon, a leading hybrid solid-state LiDAR, is being produced at scale, all while meeting the stringent automotive-grade reliability standards of the industry and the demanding technical expectations of our partners. Solid-state LiDAR just can’t deliver in the same way today. Will solid-state get there and replace hybrid? We’ll see, but for the moment, hybrid solid-state is the best LiDAR technology available and the only one deliverable at scale to meet our partners’ demanding needs.

Moving forward, we will continue to evaluate and research a wide variety of LiDAR technologies and approaches and will always be dedicated to providing our customers with the best LiDAR and sensing technologies available. We’ll continue to select the technology route that we consider the most mature, suitable, and cost-effective based on practical usage scenarios.

PB: LiDAR technology often works in conjunction with other sensor systems like cameras and radar. How do you envision the synergy between LiDAR and these complementary technologies in enabling safer and more reliable autonomous driving systems?

YL: We expect these three sensor technologies to coexist and complement each other for a long time. LiDAR technology itself plays an undeniably critical role in autonomous driving systems by providing high-precision 3D information that other sensors are unable to generate. But when paired with other sensor systems, such as cameras and radars, there are synergistic benefits to the overall safety and reliability of the system. For example, cameras provide high-resolution images for object and scene recognition. LiDAR, on the other hand, provides precise distance and 3D spatial information, detecting objects that may be challenging for cameras, such as pedestrians or obstacles in low-light conditions. However, through the collaboration of cameras and LiDAR, more comprehensive and accurate perception information can be obtained, and the surrounding environment can be perceived and understood more accurately and holistically. The result is that autonomous driving systems are able to make better and quicker decisions, leading to safer, more comfortable experiences for us, our friends, our families, and society as a whole. 

PB: Looking ahead, what do you foresee as the next major breakthrough or innovation in LiDAR technology that will have a transformative impact on the automotive industry?

YL: With LiDAR already being recognized as a necessary sensor for intelligent driving and with our Falcon sensors already in tens of thousands of cars on roads all around the world, we’re really excited about the transformative impacts that it is already enabling.

That said, we’re just getting started and think LiDAR is also. We’re actively exploring new technologies and approaches, and we’re really excited about some of the promising new technologies we’re working on. For example, while 1550nm and 905nm have been established as the de facto wavelength bands for emitted lasers, what happens when you explore higher bands? What characteristics of detected targets can be improved, and what limitations of current LiDAR systems can we mitigate? These are the kinds of questions we’re asking and the kind of research we’re conducting, all in the hopes of helping to spark the next big wave of innovations in LiDAR and sensing technology.

PB: Is there anything else you would like to add?

YL: Thank you so much. We’re excited to help bring the power of intelligent vision to everyone and everything and can’t wait to see all the possibilities that will be unlocked as a result. With LiDAR, we think the future is bright, and we’re excited to be a part of such a dynamic and growing industry.

Related articles

up