标签： 杭州品茶网

Luxin Venture Capital： Dig deep into ＂Moments＂ and introduce ＂hard technology＂

　　The ecosystem built by technology and products of technology enterprises is profoundly changing life scenes and industrial patterns. In the current era of accelerating towards intelligence, innovative technology ecology has increasingly become a key element for the transformation of old and new kinetic energy and the high-quality development of industries.

　　Technology ecosystem refreshes smart experience

　　What kind of intelligent experience can a technology ecosystem bring?

　　Driving a smart car, encountering complex traffic scenarios such as road construction obstacles, pedestrians crossing the road, and suddenly appearing electric vehicles, can identify dangers earlier and faster, easily change lanes to avoid obstacles, actively avoid risks, and actively stop.

　　Using a smart PC, you can quickly and accurately extract key information from text, audio, and video content without time and effort;

　　Open the smart screen and point to the remote control to make the control as convenient as playing with a mobile phone;

　　Wearing smart earphones, you can automatically switch the noise reduction mode according to the environment without hands, and smart watches and smart bracelets can detect health indicators and generate micro-physical examination reports;

　　Install the smart door lock, no matter the light is bright or dark, the angle and distance change can be accurately and securely unlocked by facial recognition…

　　The most important thing is that all these products can achieve synergy, such as mobile phones can be remotely parked, watches can be used as car keys, mobile phone applications can be "touched" to the car screen, smart screen, and even smart door locks can be linked with mobile phones, smart screens, and other multi-screens.

　　This is the full scenario of the Internet of Everything showcased at Huawei Hongmeng Ecological Spring Communication Conference on April 11. The underlying distributed capabilities of Hongmeng Ecological can achieve full-scene device collaboration, allowing smart experiences to pervade all aspects of people’s lives.

　　"Since January 18 this year, Huawei announced that the first batch of more than 200 application manufacturers are accelerating the development of Hongmeng native applications. By the end of March, there will be more than 4,000 applications, and the growth rate will be as high as 20 times in just two months, covering more vertical applications." Yu Chengdong, executive director of Huawei, CEO of end point BG, and chairperson of smart car solutions BU, revealed at the communication meeting that Hongmeng ecology has achieved a complete vertical industry coverage, including convenient life, travel and tourism, financial management, social information, productivity tools, corporate office, audio and video entertainment. Applications in various fields are competing to start Hongmeng native application development, constantly injecting new vitality into Hongmeng ecology.

　　Among them, it is particularly worthy of attention that Huawei, which "does not build cars", is helping more car companies to "build good cars": with more than 30 years of technical accumulation in the ICT field, it has shown obvious advantages through Hongmeng Zhixing ecology to help the commercial success of automotive industry partners. Taking the Q Jie jointly built by Huawei and Sailis as an example, the number of phenomenon-level explosive models Q Jie’s new M7 is set to reach 174,000 in 7 months; the full-scene smart flagship SUV Q Jie M9 has been listed in just three months. A total of 70,000 vehicles have been set to break through, breaking the domestic 500,000 yuan level luxury car sales speed record. In the first quarter of this year, the cumulative sales of the new M7 reached 74,400 units, winning the first place in the sales of single models of new forces in China.

　　"If China wants to build a good car, it cannot do without Huawei. If China wants to go from an automobile power to an automobile power, it needs the empowerment of Huawei Hongmeng System." Yin Tongyue, secretary of the party committee and chairperson of Chery Holding Group Co., Ltd., said at the communication meeting. When talking about the S7 smart car of Hongmeng Zhixing, which Chery has cooperated with Huawei, Yin Tongyue said that the smart world is Chery’s most important project. Chery has become an independent business unit of the Lizhijie brand, led by the R & D leader and the manufacturing leader, and the exclusive manufacturing industrial production line and special project resources are guaranteed.

　　This is Huawei’s booth at the 4th China International Consumer Goods Expo, where Hongmeng ecological products, including Hongmeng Zhixing, are on display. Photo by Yang Guanyu, a reporter from Xinhua News Agency

　　Full scene artificial intelligence deeply empowered

　　From smart travel to smart home, from smart office to smart wear, Hongmeng’s new products are packed with various innovative technologies, among which AI (artificial intelligence) is undoubtedly a high-frequency word.

　　One of the biggest highlights of the new Huawei MateBook X Pro laptop released at the Hongmeng Ecological Spring Communication Conference is its thinness, with a thickness of only 13.5 mm and a weight of only 980 grams. According to Yu Chengdong, the MateBook X Pro is the only notebook equipped with Intel Core Ultra 9 high-performance processor within 1kg, and it is also the world’s first 14.2-inch flexible OLED notebook. More notably, this is the first time that Huawei PC has applied Huawei Pangu model. Its AI summary function can quickly and accurately extract key information from text, audio and video content. In addition, the MateBook X Pro is also equipped with Huawei AI space function for the first time. Through this one-stop AI ability aggregation entrance, it can make code programming, PPT production, paper reading and typesetting, collection of learning materials, video picture creation, travel planning and other scenes more efficient and convenient. On this heavyweight AI PC, Huawei also jointly tuned with Wanxing Meow Shadow to call NPU for AI special effects creation, bringing users a smarter and more efficient experience; in cooperation with WPS, it brings a series of AI functions including text intelligent typesetting, dialogue operation tables, and brain map-to-PPT.

　　It is understood that the AI capabilities of the Pangu model have been mainly used in the industry before. This move on the PC track will be an important step in the exploration of the Pangu model on the consumer side, and the on-board smart assistant Xiaoyi will also be connected to the Huawei Pangu model in May.

　　In fact, not only is there a big model blessing, but the new products of Hongmeng Ecological all scenes are deeply empowered by AI technology. For example, behind the staggering smart driving capabilities of the Smart World S7 is Huawei’s rich scene library built on the AI training cluster, in which more than 12 million kilometers of simulation tests are carried out every day, and the learning and training computing power reaches 3.3E FLOPS (tens of billions of floating-point operations per second). The model update speed is iterated every 5 days, and the intelligent driving ability is continuously iterated and upgraded with massive data; the AI camera equipped with Huawei’s smart screen has powerful intelligent perception capabilities, which can realize 3D bone recognition, sitting posture or distance reminders, portrait tracking, etc.; Huawei Smart Door Lock Plus applies self-developed AI technology to 3D facial recognition, which not only independently learns subtle changes in the face and environment, but also has powerful AI 3D anti-counterfeiting detection, and can resist photos, videos, masks Five attack methods, such as headgear and 3D headmold, ensure that facial recognition is more convenient, safer and smarter from the bottom of the technology.

　　Accumulate new potential energy to open up more space

　　With this year’s "artificial intelligence +" being written into the government work report for the first time, artificial intelligence is expected to deeply empower the development of the real economy and become an important engine for the development of new quality productivity. At the application level, artificial intelligence is deeply integrated with all walks of life, and is expected to build a data-driven, man-machine collaboration, cross-border integration, and co-creation and sharing of intelligent economic forms, giving birth to new economic growth points.

　　Take the PC industry as an example. The industry generally believes that AI PCs will reshape users’ PC experience, profoundly change people’s work patterns and lifestyles, improve the efficiency of the whole scene, and set off a new round of PC replacement.

　　IDC, a market research firm, reports that the proportion of AI PCs installed in China’s PC market will only be 8.1% in 2023, and the proportion will increase significantly to 54.7% in 2024. It is expected to rise rapidly in the next few years, reaching 85% in 2027 and becoming the mainstream of the PC market.

　　Huaxin Securities Research Report said that the AI PC usage scenarios are highly overlapping with the application scenarios currently covered by the AI big model, and are called "the best carrier of the big model." Chip processor manufacturers such as Qualcomm, AMD, and Intel have launched processors for AI PCs.

　　Industry insiders believe that the launch of AI PC products by the top end point brand marks the arrival of the AI PC era. At present, Apple, Microsoft and other tech giants have released AI PCs. As the top three manufacturers in the domestic PC market and the only one growing against the trend, Huawei’s entry into the new AI PC track will help the recovery of the PC industry chain and is expected to add new vitality to the PC industry.

Horizon perception algorithm – a more open visual perception solution than Mobileye

　　As a leading enterprise in the ADAS industry, Mobileye currently holds more than 60% of the global market share of visual perception chips. At the same time, the closed visual perception ecosystem established by Mobileye has also become the mainstream form of the industry. Conditional open and relatively convergent perception API requirements have helped Mobileye establish standardized perception interface solutions and quickly promote products to the world, attacking cities and territories.

　　However, time has changed. As current visual ADAS capabilities continue to be upgraded and iterated, the industry expects more open solutions to help OEMs and Tier1 create differentiated and more adaptable ADAS capabilities.

　　The complexity and locality of driving scenarios call for open solutions

　　Once upon a time, a domestic OEM tried to develop an autonomous parking solution based on Mobileye’s visual perception solution. During the implementation of the project, it was necessary to identify the railing at the entrance of the parking lot. However, Mobileye’s closed solution did not support customers to update the perception algorithm independently, resulting in development difficulties.

　　This reflects the dilemma faced by OEMs. On the one hand, Mobileye’s visual perception solution is still the most mature solution at present, with many advantages in terms of product maturity, project risk control and quality management, which OEMs cannot refuse. On the whole, it is a situation of more than conservative and less aggressive. But on the other hand, this closed solution is like a black box. Because of the inability to differentiate and customize development, the performance is homogenized; moreover, in localized driving scenarios, it cannot fully meet the requirements. This limits OEMs to make more aggressive product solutions, which will reduce product competitiveness in the long run.

　　The particularity of China’s driving scene requires a localized sensing solution to meet the functional requirements of ADAS/autonomous driving.

　　At the same time, in order to meet the high reliability requirements of ADAS/autonomous driving for perception, more open perception solutions are also required. For example, if we can identify richer targets and more types of semantic segmentation, then we can obtain richer semantic information, so that different categories can be cross-validated. For example, road shoulders and sidewalks have a clear auxiliary verification function for the judgment of the drivable area. Fixed targets on the roadside, such as traffic signs and street lamps, are of great help in positioning.

　　If the road is covered with snow on a snowy day, how can you tell where the shoulder is? The movie "If You Are the One" describes a dialogue plot:

　　"Hey, you’ve been to [Hokkaido], you must know what the arrows on both sides of the road are for?" Answer: "Oh, in winter, the snow in Hokkaido is very thick, and the arrows instruct people not to drive out of the shoulder of the road."

　　If we can recognize the arrows on either side of the highway, we can reason about the boundaries of the drivable area. The development trend of perception technology requires more open solutions

　　The development of perception technology can be summarized as follows:

　　From simple scenes to complex scenes

　　From high-frequency targets to general targets

　　From 2D perception to 3D perception

　　From reality-oriented perception to prediction-oriented perception

　　All of these trends have further increased the richness and complexity of perception, making it impossible for vendors to use a standard perception solution to meet their needs.

　　In the era of software-defined cars, automakers need a more open approach

　　Intelligence is the core element of brand differentiation in the future, mainly through the addition of software functions. The post-deployment of software will be the general trend, which means that most software functions will be delivered after the car leaves the factory, and software iteration OTA will be the new normal. This trend is particularly important for mobility service operators. Service requirements in various scenarios need to be based on the functions of existing fleets and met by continuous upgrades and iterations.

　　In the future, the car delivered by OEMs will not be a product with solidified functions, but a robot that continues to evolve. Throughout the life cycle of the car, the hardware platform can continue to support software iterative upgrades. The efficiency of software development and differentiated functions will determine the success or failure of this intelligent competition.

　　In order to improve the efficiency of software development, from the perspective of system architecture, service-oriented system architecture (SOA) will become the mainstream, which requires the creation of a new perception solution that meets the requirements of four aspects: a highly open, consistent and complete tool chain, a strong computing power reserve, and strong scalability to meet the requirements of different levels of vehicle platforms.

　　Open perception intermediate results help domestic ADAS functions continue to evolve

　　At present, the perception algorithm API is still relatively closed in the industry. Many features are difficult to implement (such as the augmented reality display function of ADAS function in the infotainment domain), in part because the perception algorithm is only provided for internal use of the intelligent forward-looking camera module and is unwilling to be provided to other subsystems. And the Horizon perception intermediate result can be fully open. And because the low-level semantics of the Horizon algorithm are very rich, the fully open perception intermediate result can support customers to develop more complex functions at the application layer.

　　So far, the Horizon algorithm can support 10 types of dynamic targets and 53 types of static targets. Among them, dynamic targets include: adults, children, cyclists and other pedestrians, as well as cars, SUVs, vans, trucks, passenger cars, motorcycles and elderly scooters 7 categories; while static targets include 8 categories of lane lines, 2 types of traffic lights and 43 types of traffic signs. The richer perception information than Mobileye provides a solid perception foundation for customers to realize differentiated functions.

Rich Perceptual Information Provided by Horizon Algorithm

　　Open and comprehensive toolchain, practicing the concept of "deeply empowering"

　　In order to increase the agility of visual algorithm iteration and better support various extreme perceptual scenarios in China, Horizon has launched the AI chip tool chain Horizon OpenExplorer (Horizon "Tiangong Kaiwu"), including data, training and device deployment tools, such as model training tools, inspection and verification tools, compilers, emulators, embedded development kits, etc. (Figure 9), forming a closed loop. Data generates models, which can be deployed to devices for operation, and can guide the tuning of models during operation, and even collect new data. Such a self-evolving development model can improve development speed, lower development thresholds, and ensure development quality. Based on this model, development manpower can be reduced by about 30%, development time can be saved by 50%, and more importantly, because the development threshold is lowered, the scale of developers can even be expanded by an order of magnitude.

　　Horizon will continue to upgrade the power builder to provide customers with a semi-automated processing flow. Mainly include: closed-loop iteration between data tools and models, models and on-end devices; rich model/system reference prototypes, simple and easy to use intuitive and convenient interaction means; standardized development process, plus continuous testing, integration, and deployment mechanisms.

Horizon "Tiangong Kaiwu" chip tool chain

　　More specifically, Horizon’s model training tool can support mainstream deep learning frameworks such as TensorFlow to help users train their own models; the compiler supports converting the open-source training framework model format to the binary format on the chip; and the embedded development kit can support customers to call the algorithm library to develop their own applications, allowing customers to quickly deploy applications in chip manufacturers. The entire toolchain package can cover the complete development link (Figure 10). After optimization by the Horizon compiler, the memory access bottleneck of the algorithm can be greatly alleviated and the computing efficiency of the chip can be improved (Figure 11).

　　Development Process Based on "Tiangong Kaiwu" Tool Chain

Algorithms automatically optimized by the Horizon compiler provide significant performance improvements

　　Recently, Horizon partnered with South Korea’s SK Telecom to develop a dynamic crowdsourced high-definition mapping solution. SK used Horizon’s toolchain to develop a visual perception algorithm for Korean roads, demonstrating the ease of use and reliability of the toolchain.

　　In short, the opening of Horizon is a deep and multi-dimensional comprehensive opening from providing system reference solutions, to fully open sensing results, and then to full-stack solutions of toolchains, fully empowering the intelligent development of the automotive industry. If customers advocate the concept of division of labor and cooperation to maximize efficiency, Horizon provides software and hardware integrated chip solutions; if customers prefer to maximize capabilities and want to use their own algorithms, Horizon provides customers with pure chips and a complete toolchain to help customers achieve deep enough development freedom and practice the long-term commitment of "deep empowerment".

Development Status of Artificial Intelligence AI Industry in China

　 　 　1. Development status of AI

　 　 With the advent of the fourth industrial revolution, artificial intelligence has gradually entered reality from science fiction. Since the concept of artificial intelligence was first put forward in 1956, the development of artificial intelligence has experienced ups and downs. With the breakthrough of core algorithms, the rapid improvement of computing power, and the support of massive Internet data, artificial intelligence finally ushered in a qualitative leap in the second decade of the 21st century, and became the focus of science and technology attracting worldwide attention.
　
　 　 For China, the development of artificial intelligence is a historic strategic opportunity, which is crucial to alleviate the pressure of population aging in the future, meet the challenges of sustainable development and promote the transformation and upgrading of economic structure. As the core driving force of the new round of industrial transformation and the strategic technology leading the future development, the state attaches great importance to the development of artificial intelligence industry. In 2017, the State Council released the "New Generation Artificial Intelligence Development Plan" to strategically deploy the artificial intelligence industry; In the government work reports in March 2018 and March 2019, it was emphasized that it is necessary to accelerate the development of emerging industries and promote research and development of artificial intelligence.app; applicationCultivate a new generation of information technology and other emerging industrial clusters to expand the digital economy.
　
　 　 Since the deep learning algorithm was put forward in 2006, the application of artificial intelligence technology has made a breakthrough. Since 2012, the explosive growth of data has provided sufficient "nourishment" for artificial intelligence, and deep learning algorithms have achieved breakthroughs in speech and visual recognition, making it possible for the artificial intelligence industry to land and commercialize.
　
　 　 2. Development layout of AI


　 　 On the whole, China’s artificial intelligence industry has formed a pattern of coordinated development in Beijing-Tianjin-Hebei region, Yangtze River Delta region and Pearl River Delta region, especially in the eastern coastal areas. According to IDC’s Evaluation Report on the Development of Artificial Intelligence Computing Power in China in 2019-2020, the TOP5 cities in China’s list of artificial intelligence cities in 2019 are Beijing, Hangzhou, Shenzhen, Shanghai and Guangzhou in turn.
　
　 　 Among them, Beijing has a relatively mature layout and a relatively complete industrial chain. In particular, Zhongguancun Science City and other institutions provide favorable policies, technologies and talents for the development of Beijing’s artificial intelligence industry. At present, the number of enterprises with artificial intelligence in Beijing ranks in the forefront of the country, about 300.
　
　 　 The number of artificial intelligence enterprises in Shanghai is less than that in Beijing, but it is also ahead of other provinces and cities. Shanghai plans to build an artificial intelligence highland and create a first-class artificial intelligence innovation ecology. At present, it has laid out a "human-shaped" spatial structure, which is distributed in 8 regions and 11 industries. In addition to Shanghai, Jiangsu, Zhejiang and other places are also actively deployed, and artificial intelligence enterprises are more concentrated.
　
　 　 Guangdong’s scientific and technological innovation ability is also in a leading position, especially the planning and development of Guangdong-Hong Kong-Macao Greater Bay Area attaches importance to the layout of artificial intelligence. Up to now, Guangdong Province has successfully identified two batches of 8 parks as "Guangdong Artificial Intelligence Industrial Park". In addition, relying on the important opportunities of national characteristic towns, thousands of enterprises and thousands of towns projects, the Pearl River Delta National Independent Innovation Demonstration Zone and the Guangzhou-Shenzhen Science and Technology Innovation Corridor, Guangdong encourages local cities to speed up the application layout of artificial intelligence industry and build a number of artificial intelligence towns.
　
　 　 Hangzhou attaches great importance to the basic research of artificial intelligence. The "Xiangyun" DRMS system is the first data resource platform based on artificial intelligence in China. It is a full-volume, real-time "data warehouse" of Hangzhou’s urban data brain, which can provide functions such as real-time data collection, data Rubik’s cube and data governance, and creatively apply artificial intelligence to data exchange, sharing and opening. Driven by leading enterprises such as Alibaba, Netease, zhejiang dahua, Hikvision and Straight Flush, Hangzhou has formed a number of distinctive applications in smart cities, finance and other fields. In addition, Hangzhou actively builds an artificial intelligence industrial cluster, taking Hangzhou High-tech Industrial Development Zone and Hangzhou Chengxi Science and Technology Innovation Corridor as the core, optimizing the layout of the city’s artificial intelligence industrial structure and realizing orderly competition and dislocation development.
　
　 　 3. Market demand and scale of AI
　
　 　 China’s artificial intelligence industry focuses on diversified application scenarios. Under the guidance of China’s national conditions and market demand, it aims at the practical needs of intelligent transformation and upgrading in transportation, medical care, finance, security and other fields, and focuses on selecting one or several key areas for key layout, focusing on the whole life cycle of big data in the industry, and taking the lead in promoting commercialization by optimizing scene design. In recent years, the competitive situation of differentiation and regionalization has attracted a large number of emerging artificial intelligence enterprises in China, which has promoted the continuous outbreak of China’s industrial scale.
　
　 　 According to the statistics of Gaozhan think tank, the scale of China’s artificial intelligence industry reached 8.31 billion US dollars in 2018, and by 2019, the industrial scale exceeded 10 billion US dollars, reaching 10.55 billion US dollars. Among them, the industrial scale of AI basic layer, technology layer and application layer reached 1.99 billion US dollars, 3.06 billion US dollars and 5.5 billion US dollars respectively. Driven by policies and markets to accelerate the implementation of a new generation of artificial intelligence application scenarios, it is estimated that the industrial scale will approach 30 billion US dollars in 2022.
　
　 　 Chart: The scale of artificial intelligence industry in China from 2015 to 2019 (USD 100 million)
　 　 Source: Gaozhan think tank
　
　 　 4. Number of AI enterprises
　
　 　 According to the statistics of Gaozhan think tank, by the end of 2019, there were 779 artificial intelligence enterprises in China, accounting for 21.9% of the total number of artificial intelligence enterprises in the world, second only to the United States. In terms of the distribution of start-up time, the establishment time of artificial intelligence enterprises in China concentrated in 2010-2016, accounting for 67.3%, and reached its peak in 2015, and the number of enterprises established in 2016-2019 gradually declined.
　
　 　 Chart: Time distribution of artificial intelligence enterprises in China as of 2019.
　 　 Source: Gaozhan think tank
　
　 　 In terms of geographical distribution of enterprises, artificial intelligence enterprises in China are mainly concentrated in the three metropolitan areas of Beijing-Tianjin-Hebei, Yangtze River Delta and Pearl River Delta, accounting for 44.8%, 28.7% and 16.9% respectively. Among the provinces and cities, artificial intelligence enterprises are mainly distributed in Beijing, Guangdong, Shanghai, Zhejiang, Jiangsu and other places, accounting for 88.4% of the total.
　
　 　 Chart: Regional distribution of artificial intelligence enterprises in China in 2019
　 　 Source: Gaozhan think tank
　
　 　 5. Market structure and market share of AI enterprises
　
　 　 The artificial intelligence application market mainly includes three aspects: computer vision application, speech semantic application and machine learning algorithm platform. Among them, computer vision application scenarios are rich and have great commercial value, and about 40% of artificial intelligence enterprises are concentrated in the field of visual artificial intelligence. According to the statistics of Gaozhan think tank, in 2019, the largest manufacturers in China’s artificial intelligence computer vision application field included Shangtang Technology, Defiance Technology, Yuncong Technology, Yitu Technology, Hikvision, Innovative Qizhi, Smart Eye Technology and Dahua Co., Ltd., accounting for 79.5% in total; The manufacturers occupying the largest market in the field of artificial intelligence speech semantic application include Iflytek, Tulsi and Xiao I.robot, Baidu Net News, Jietong Huasheng, Yunzhisheng, Shenzhou Taiyue, Alibaba Cloud and Spirits, accounting for 39.2% in total; The leaders of artificial intelligence algorithm learning platform include Fourth Paradigm, Alibaba Cloud, Baidu, Tencent Cloud, AWS and Microsoft. The main participants include Merrill Lynch Data, Jinshan Cloud, Jiuzhang Yunji, Intelligence Explorer, Xinghuan Technology, Xinhua III and Neusoft.
　
　 　 Chart: China artificial intelligence market pattern in 2019