Follow “Youzhiliao” for valuable insights!
1. Introduction to Brain-like Complementary Vision Chip
Core Definition: The brain-like complementary vision chip is a new type of sensor designed inspired by biological visual systems. It breaks through the limitations of traditional CMOS image sensors’ “frames” and adopts an “event-driven” working mechanism. Its core innovation lies in the fusion of dynamic, asynchronous perception capabilities based on “event cameras” with static, high-resolution texture perception capabilities based on “RGB cameras” through chip-level heterogeneous integration technology, thereby achieving the complementary functionality of two visual modalities on a single chip, simulating the dual-path processing mechanism of the human eye’s “retina-visual cortex”.
· Core Technology:
· Heterogeneous Integration: “Integrating event cameras with RGB pixels heterogeneously” is the essence of this chip. It is not simply packaging two chips together, but rather arranging or stacking event-driven pixels and traditional RGB pixels at the wafer level to achieve complementary raw data at the perception level.
· Biomimetic Photoreceptors: “1.28 million biomimetic photoreceptors” refers to the total number of pixel units integrated on the chip, which can mimic the photoreceptor cells in the human eye’s retina, responding to changes in illumination (events) or absolute brightness (RGB).
· Performance Indicators:
· Power Consumption 3%: The event-driven characteristic means almost zero power consumption in static scenes, with data output only when brightness changes occur at pixel points, resulting in overall low power consumption. Compared to traditional cameras that continuously output full-frame data, a reduction to 3% power consumption is reasonable.
· Low Light Recognition Rate 99.3%: The sensitivity of the event camera to brightness changes allows it to work effectively even in extremely low light environments; this data should be supported by a specific test set.
· Foggy Day 1000km Road Test: This is a milestone verification project planned for the chip, aimed at proving its reliability and superiority in extremely harsh environments (such as fog, strong light, and high speed), which is a key step towards automotive applications.
· Source: The initial report of this breakthrough should originate from the official website of Peking University, the School of Information Science and Technology, or the Institute of Artificial Intelligence.
Research teams (such as Professor Huang Tiejun’s group) have published academic papers on technical details in top international conferences or journals (such as ISSCC, VLSI Symposium, Nature Electronics, IEEE JSSC). Before the formal publication of the papers, one can follow their preprints published on arXiv.org.
SMIC (Semiconductor Manufacturing International Corporation), as the chip manufacturer, may mention this collaborative project in its technical forums or official news, confirming the advancement of its manufacturing processes (such as 55nm or 40nm specialty processes).
2. Comparative Analysis with International Standards and Its Significant Implications
International Comparison:
International leading levels (Sony, Samsung, Prophesee, iniVation), domestic levels (represented by Peking University-SMIC’s “Lingmou-1”)
Technical Route:
Internationally, the discrete route is dominant: Sony/Samsung lead the high-end CIS market; Prophesee (France) and iniVation (Switzerland) focus on event cameras, followed by algorithmic post-fusion.
Domestically, the fusion route is leading: pioneering the native heterogeneous integration of event and RGB pixels at the chip level, providing integrated raw data of “sensing” and “knowing”.
From “following” to “creating”. This avoids information loss and computational waste in post-fusion, achieving architectural-level innovation.
Industrialization:
Internationally, Sony and Samsung dominate the consumer electronics CIS market; Prophesee collaborates with Sony and Qualcomm to provide event vision solutions.
Domestically, it is the first to propose and realize a complete technology chain for complementary vision chips (design-manufacturing-verification), targeting autonomous driving as the ultimate application scenario.
International giants have advantages in traditional markets and ecosystems; China has seized a first-mover advantage and definition rights in new tracks with new architectures.
Performance:
Internationally, traditional CIS has inherent defects in high-speed and high-dynamic range scenarios. Pure event cameras lack static texture information.
Domestically, comprehensive performance achieves generational advantages in low light, high speed, and high dynamic range (HDR) scenarios, surpassing human night vision capabilities at the perception level for the first time.
This addresses the pain points of traditional visual systems, achieving breakthroughs in performance indicators and overall leadership.
Significant Implications:
1. Architectural Innovation: This is not a simple process improvement, but rather draws inspiration from biological mechanisms, creating a new technological paradigm of “complementary vision,” providing the possibility for China to overtake in the field of intelligent visual sensing.
2. Addressing Industrial Pain Points: Directly targeting the perception bottlenecks of core applications such as autonomous driving and robotics in extreme weather and high-speed scenarios, providing unprecedented solutions, and is expected to become a standard sensor for L4-level and above autonomous driving.
3. Autonomous Control: From chip design, manufacturing (in collaboration with SMIC) to application verification, forming a complete domestic technology closed loop, breaking the monopoly of Europe, America, Japan in the high-end visual sensing field.
4. Empowering AI: The provided “event stream + RGB” native fused data aligns more with the processing mode of spiking neural networks (SNN), potentially promoting the development of the next generation of low-power, high-performance brain-like AI.
3. Application Scenarios and Market Space
Application Scenarios:
1. High-level Autonomous Driving (ADAS/AD): Core application scenario. Solving perception challenges in fog, rain, snow, night, and strong backlight, enhancing safety.
2. Consumer Electronics: Low-power, high-dynamic visual perception for mobile phones and AR/VR glasses, achieving Always-on Vision applications.
3. Industrial Inspection: Precise detection of tiny defects and rapidly moving objects on high-speed production lines.
4. Robotics: Endowing robots with more agile visual response capabilities and environmental understanding.
5. Security Monitoring: Achieving intelligent monitoring without blurriness in extremely low light conditions.
Market Space:
· Core Driving Market – Autonomous Driving: According to market research institutions like Yole Group, the automotive camera market is expected to exceed $10 billion by 2028. The brain-like vision chip, as the next generation of high-performance sensors, corresponds to a potential market space of billions of dollars, even with low initial penetration rates.
· Long-term Imagination Space: If its low-power, high-performance advantages are applied in mobile phones, IoT, etc., the market ceiling will be further opened.
· Current Stage: The market is on the eve of industrialization from 0 to 1. The 1000km road test of “Lingmou-1” is a key step towards automotive-grade (AEC-Q100) certification and mass production. Investment should focus on its industrialization progress and the landing situation of the first batch of customers.
4. Relevant Policies in the 14th Five-Year Plan to the 15th Five-Year Plan
· “14th Five-Year Plan Outline”: Clearly lists “brain science and brain-like research” as a major national science and technology project, emphasizing the need to “focus on artificial intelligence… and other frontier fields,” and “key core technologies such as new generation artificial intelligence key algorithms and intelligent sensors.”
· National AI Innovation Application Pilot Zone: In the policies of pilot zones such as Beijing, Shanghai, and Shenzhen, innovative technologies in intelligent sensing, autonomous driving, and other fields are given key support.
· “14th Five-Year Plan for Robot Industry Development”: Emphasizes the need to focus on breakthroughs in key technologies such as “intelligent sensors.”
· 15th Five-Year Plan Outlook: It is expected that during the 15th Five-Year Plan period, support for “brain science and brain-like research” will shift from “scientific research” to “technology transfer” and “industry cultivation.” Major original innovative achievements like “Lingmou” will become key projects supported by the state to build industrial chains.
Source: Official websites of the People’s Republic of China Central Government, Ministry of Industry and Information Technology, Ministry of Science and Technology.
5. Research on Listed Companies and Leading Enterprises in the Industrial Chain Cooperation
This field is at the forefront, and there are few directly related listed companies; attention should be paid to enterprises with close cooperation with core technology teams.
SMIC 688981.SH/0981.HK: As the chip manufacturer of “Lingmou-1,” its role is crucial. This chip requires special wafer processes to achieve heterogeneous integration of event and RGB pixels. Source: Official press releases from both parties (Peking University, SMIC). Core manufacturing platform. It proves its special process R&D capabilities and strength in supporting cutting-edge innovations.
Weir Shares 603501.SH: Its subsidiary OmniVision is a global leading CIS supplier and has publicly released event vision sensors. Although it has not directly mentioned cooperation with “Lingmou,” as an industry leader, its technological layout and follow-up speed are worth close attention. Source: Company annual reports, investor relations activity records. Traditional CIS leader, potential competitor or collaborator. It has strong market channels and automotive customer resources.
Note: The intellectual property (IP) of the “Lingmou” chip belongs to Peking University. Its industrialization path is most likely to establish a new startup company for technology transfer and market promotion (similar to the establishment model of “Cambricon”). This potential “Peking University system” startup is the most important investment tracking target in the future. Currently, no listed company directly holds its equity.
6. Risk Warning
1. Industrialization Risk: There is a huge engineering gap to cross from stunning samples in the lab to stable, reliable, low-cost automotive-grade mass production products, including yield, consistency, and reliability certification.
2. Technical Route Competition Risk: Traditional CIS giants (Sony, Samsung) and event camera startups (such as Prophesee) are also rapidly iterating and may launch similar fusion solutions, leading to intense competition.
3. Ecosystem Construction Risk: New sensors require supporting algorithms, software toolchains, and processor support; building a complete developer ecosystem requires time and significant investment.
4. Market Acceptance Risk: Autonomous driving companies are cautious about adopting new sensors and require long-term validation before large-scale procurement.
5. Valuation Risk: Related concepts may be speculated by the market, and caution is needed regarding excessively high valuations that deviate from industrialization progress.
Conclusion and Investment Recommendations
The Peking University-SMIC “Lingmou-1” brain-like complementary vision chip is a world-class original innovation in China’s intelligent sensing field, with significant implications for creating a new technological direction and achieving a breakthrough in key performance indicators over international levels for the first time.
Investment Logic:
1. Cutting-edge Technology Investment: It belongs to high-risk, high-potential return cutting-edge technology layouts, suitable for risk-tolerant investors.
2. Industrial Chain Sequence: The core is the design and manufacturing of the chip, followed by system integration and application in scenarios such as autonomous driving.
3. Core Targets:
· Short-term focus on SMIC (688981.SH), as its manufacturing platform will continue to benefit from domestic high-end chip design orders.
· Closely monitor the financing and progress of related startups derived from Peking University, as this is key to capturing core value.
· Long-term focus on Weir Shares (603501.SH) and other industry leaders’ follow-up and layout in this technology route.
4. Tracking Points:
· The official results release of the “Lingmou-1” 1000km foggy road test.
· The establishment and financing dynamics of the R&D team’s industrialization company.
· Whether mainstream autonomous driving manufacturers (such as BYD, NIO, Huawei, etc.) announce cooperation or testing with it.
Investors are advised to deeply understand the revolutionary potential of this technology while remaining rational and closely tracking every substantial step of its industrialization.
Disclaimer: This report is compiled, verified, and analyzed based on publicly available information, aiming to provide objective research references and does not constitute any investment advice. Investment carries risks; proceed with caution.