The Action Plan sets forth four major tasks.
First, focus on nurturing and developing intelligent products such as smart car, intelligent service robot, intelligent unmanned aerial vehicle, medical imaging aided diagnosis system, video image identification system, intelligent voice interactive system, intelligent translation system, smart home product, etc. Intelligent products in the economic and social integration.
The second is to focus on the development of intelligent sensors, neural network chips, open source open platform and other key areas, tamping the development of artificial intelligence industry hardware and software foundation.
Third, to deepen the development of intelligent manufacturing, to encourage the exploration and application of the new generation of artificial intelligence technology in all aspects of the industrial field, to enhance the innovation capability of intelligent manufacturing of key technology and equipment, and to foster the promotion of a new mode of intelligent manufacturing.
The fourth is to build a public support system for industry training resources, standard testing and intellectual property service platform, intelligent network infrastructure, network security and other industries, and improve the environment for the development of artificial intelligence.
Focus on cultivating eight categories of smart products
In developing smart products, the Plan of Action sets out detailed development goals:
By 2020, JLV will establish a smart, smart and real-time smart car network intelligence platform to form a platform-related standard that supports highly autonomous driving (HA level).
In terms of intelligent service robots, breakthroughs will be made in key technologies such as environment awareness, natural interaction, autonomous learning and human-computer collaboration by 2020. The intelligent home service robot and intelligent public service robot achieve mass production and application, medical rehabilitation, , Fire rescue and other robots to achieve prototype production, complete technical and functional verification, to achieve more than 20 application demonstration.
In the field of intelligent UAVs, by 2020, the smart consumer UAV 3-axis mechanical stabilization head reaches 0.005 degrees, realizing 360-degree omnidirectional perception of obstacle avoidance and realizing automatic and intelligent forced avoidance of air traffic control areas.
In the field of medical imaging aided diagnosis system, by 2020, the advanced multimodality medical imaging diagnosis system in our country has the detection rate of more than 95%, the false negative rate less than 1% and the false positive rate less than 5%.
In the video image identification system, the effective detection rate of face recognition in complex dynamic scenes exceeds 97% by 2020, and the correct recognition rate exceeds 90%, which supports the recognition of face features in different regions.
In terms of intelligent voice interactive system, by 2020, the average accuracy rate of Chinese speech recognition in multi-scenarios will reach 96%, the recognition rate of 5 meters in far field will exceed 92%, and the accuracy of user dialogue intention recognition will exceed 90%.
In the field of intelligent translation systems, the multilingual intelligent mutual translation achieved significant breakthroughs by 2020, with the accuracy of translation of products in Chinese-English translation and English-Chinese translation exceeding 85%, and the accuracy rate of intelligent translation between minority languages and Chinese languages significantly improved.
In the field of smart home, by 2020, the category of smart home products will be significantly enriched, the penetration rate of smart TV market will reach over 90%, and the intelligent product level of security products will increase significantly.
In addition, the document also encourages local governments to step up investment in fostering a group of leading enterprises in artificial intelligence, building an artificial intelligence industrial cluster, and guiding multi-lateral capital to participate in the development of artificial intelligence industry so as to promote vocational schools to cultivate urgently needed skilled personnel.
Three key breakthroughs in the core of the foundation
"Action Plan" pointed out that accelerating the development and application of high-precision, low-cost smart sensors to break through the cloud-based training and terminal applications for neural network chip and supporting tools, support for the development of artificial intelligence framework, algorithm library, toolset research and development, support Open-source and open platform construction, and actively layout intelligent software designed for application of artificial intelligence to reinforce the hardware and software foundation for the development of artificial intelligence industry, focusing on the following breakthroughs in the following fields:
Intelligent Sensor Support the research and development of key technologies such as miniaturization and reliability design, precision manufacturing, integrated development tools and embedded algorithms to support the research, development and application of smart sensors based on new requirements, new materials, new processes and new principles. , Piezoelectric sensors, magnetic sensors, infrared sensors, gas sensors and other performance significantly improved, in the simulation, design, MEMS technology, packaging and personalized testing technology to achieve the international advanced level, with mobile wearable, the Internet, cars Electronics and other key areas of system design capabilities.
Development of high performance, scalable, low power, cloud-based NES chip for machine learning training applications, development of low-power, high-performance terminal neural network chip for end-use applications for machine learning computing, development And neural network chip supporting compiler, driver software, development environment and other industrial support tools. By 2020, made breakthroughs in neural network chip technology, the introduction of one or several mainstream neural network algorithm; intelligent terminals, automatic driving, Intelligent security, smart home and other key areas to achieve large-scale neural network chip business.
Open platform for open source.For the common technologies such as machine learning, pattern recognition, intelligent semantic understanding and the application of key industries such as autopilot, it supports the research and development of development framework, algorithm library and toolset for cloud training and terminal implementation, supports the open source development platform, Open up technological networks and open source communities and encourage the construction of an open computing service platform that meets the needs of complex training and encourage key leading enterprises to build a new industrial ecosystem based on open source and open technologies for software, hardware, data and application collaboration. By 2020, cloud-oriented training Open source development platform to support large-scale distributed cluster, a variety of hardware platforms, a variety of algorithms, open source development platform for terminal implementation with lightweight, modular and reliability features.