The rise of artificial intelligence is not a coincidence. It is largely due to the public’s awareness of AI technology following the victory of Alpha Go over human champions, as well as breakthroughs in technology and improvements in computational power. The three components—algorithms, data, and computing power—are indispensable. Artificial intelligence has extremely high demands for processing capability and real-time performance, and the rise of edge computing is closely related to the Internet of Things and embedded systems.
In such an era, how should we view the impending storm? The application of artificial intelligence is becoming more widespread, but which niche markets should the Internet of Things and embedded industries focus on? How should embedded and IoT researchers and engineers seize the significant opportunities brought by artificial intelligence?
Recently, the 23rd meeting of the Embedded Systems Association invited well-known experts and scholars in the industry to discuss these issues in depth! This meeting was hosted by the Embedded Systems Association and the Shenzhen Beihang University Emerging Industry Technology Research Institute, with co-hosts including the Shenzhen Beihang University IoT Research Institute, IoT Coffee, and several other organizations, under the theme “Artificial Intelligence Empowering the Development of the IoT Industry”.
The meeting consisted of two parts. The morning session was an academic seminar, chaired by Secretary-General He Xiaoqing of the Embedded Systems Association, with a welcoming speech by Director Wang Baojiang of the Beihang Emerging Industry Technology Research Institute introducing the institute’s situation.
Director Wang Baojiang of the Beihang Emerging Industry Technology Research Institute
In the era of resource networks, embedded systems will have great prospects
Professor He Limin from Beihang University
Professor He Limin, editor-in-chief of the journal “Microcontroller and Embedded Systems Applications”, introduced unbiased artificial intelligence and the future of resource networks. He believes that looking at blockchain from the Internet perspective, apart from Bitcoin, it is a decentralized, unmanned social resource trading network, which can be seen as the third revolution in the network era following the Internet and the Internet of Things. The ultimate goal of the Internet is to connect “things” and “resources” to the network, especially emphasizing that embedded practitioners should understand blockchain, as embedded systems have significant roles in the future applications of blockchain.
Embedded systems are intelligent IoT systems that, once they generally have network access capabilities, will transform the Internet into the Internet of Things. The IoT’s connection genes exist within embedded systems. As blockchain technology becomes a hot topic, embedded system technology can help simulate “resources” into digital blocks, thus facilitating the transformation from IoT to blockchain.
Blockchain will involve technologies such as digital technology, artificial intelligence, cryptography, and the construction of integrity systems, but the core of blockchain is the construction of decentralized, fair, and just smart contracts. Therefore, the realization of “unbiased” artificial intelligence is even more urgent. This is an independent evaluation system developed based on artificial intelligence that can judge fairness. Professor He emphasized that in March 2018, IBM predicted that three out of the five technologies that would have the greatest impact globally within five years were related to blockchain: encrypted anchor technology, lattice cryptography systems, and unbiased artificial intelligence. Moreover, the “encrypted anchor” chip (1 square millimeter, priced at ten cents) is about to be launched, which is considered the starting point for developing IoT blockchain solutions.
Professor He also predicted that in the next twenty years, the chaotic landscape of the market economy will undergo a complete transformation, ultimately completing the shift from an information network to a physical network and a resource network, which is precisely the era where embedded systems will thrive.
In AI, changing the perspective
Professor Kang Yimei from Beihang University
Professor Kang Yimei referred to AI as a high-potential technology, which will trigger transformations far beyond those brought by the Internet, as this technology can cover new application scenarios and its irreplaceability in these new scenarios. By analyzing the development status of domestic Internet giants like Baidu, Tencent, Alibaba, and Xiaomi in this field, Professor Kang believes that AI development will shift from the cloud to the edge, and will explode at the terminal devices, concluding that all hardware will be AI-enabled in the future.
Through analyzing the distribution of artificial intelligence technology, it was found that machine vision applications account for 42% of the entire AI market, and this number is expected to grow. However, Professor Kang discovered in his research that there is more focus on the concept and technology of AI than on products, as the transformation of AI technology into AI products requires comprehensive capabilities in algorithm innovation, software and hardware development, and engineering/productization. Professor Kang introduced that the requirements for AI algorithms in embedded systems are quite strict; they require good real-time performance in terms of time and must have small storage space in terms of space.
Intelligent systems require a good optimization system design framework
Professor Chen Yixiang from East China Normal University
Professor Chen Yixiang introduced that ubiquitous intelligent systems integrate artificial intelligence into hardware, allowing machines to make judgments, decisions, and controls in certain ways. It is a deep integration system of embedded technology, software engineering, and artificial intelligence knowledge. Professor Chen introduced that to design a good intelligent system, a good optimization system design framework must be established, which requires following system specifications, dividing hardware and software, completing hardware design, interface design, software design, and system integration steps, and detailed specificities of normative modeling and collaborative optimization design. Finally, using traffic sign recognition systems and automatic parking systems as examples, he detailed the complete design process of an intelligent system. If engineers follow the steps introduced by Professor Chen to design intelligent systems, they will avoid many detours.
Professor Huang Kai from Sun Yat-sen University introduced the situation of their UAV Research Institute and research achievements in retinal injection surgical robots.
Professor Huang Kai from Sun Yat-sen University
Afternoon entrepreneur forum activities were hosted by Executive Dean Cai Gaosheng of the Shenzhen Beihang University IoT Research Institute.
Executive Dean Cai Gaosheng of the Shenzhen Beihang University IoT Research Institute
IoT operating systems and edge computing are new trends in the development of embedded technology and IoT
Secretary-General He Xiaoqing of the Embedded Systems Association
Secretary-General He Xiaoqing first recalled the themes of the Embedded Systems Association’s activities over the past nine years. Since its establishment, the association has continuously followed the development trends of embedded systems for discussion. As of 2018, Secretary-General He Xiaoqing introduced that the wave of technological innovation has evolved from cloud computing and IoT to edge computing and artificial intelligence, and stated that the three cornerstones of embedded technology for building a future intelligent society are: to develop efficient, reliable, and low-power networking technologies; to develop green and environmentally friendly high-efficiency embedded computing technologies; and to ensure embedded functional safety and system information security will be very important.
In recent years, mergers and acquisitions in the semiconductor industry have been very frequent. Secretary-General He Xiaoqing believes that this will have a very important impact on the development of embedded systems: mergers and acquisitions will promote mainstream embedded processing to enter the IoT and AI fields; new processing technologies and business models will emerge in IoT edge computing and sensor nodes; and edge computing will play a very important role in enhancing the computing power and real-time performance of embedded AI devices. Secretary-General He Xiaoqing also detailed the use cases and current progress of edge technology.
The booming development of the Internet of Things cannot be separated from IoT operating systems. Currently, IoT operating systems are emerging, and Secretary-General He Xiaoqing introduced that unlike traditional embedded operating systems, IoT operating systems leverage cloud platforms from the outset, bringing new benefits and business models to IoT applications. He also detailed several popular IoT operating systems, such as ARM mbed OS, Huawei LiteOS, and Amazon FreeRTOS, which evolved from open-source RTOS. Finally, he emphasized that IoT security is of utmost importance and requires the joint efforts of the ecosystem!
Using FPGA to accelerate the deployment of embedded AI systems
Yao Yuan, CEO of Beijing Vision Rui Technology Co., Ltd.
Dr. Yao Yuan introduced that Beijing Vision Rui Technology Co., Ltd. focuses on fields such as wireless communication spectrum security, artificial intelligence visual perception, and nuclear and biochemical monitoring and protection, all based on deep learning computing platforms. The core of deep learning includes: the design of learning algorithms, that is, whether the designed brain is smart enough; high-performance computing capabilities, which require training a large network; substantial data support; appropriate user scenarios; and a mature business model.
Currently, deep learning computing platforms can be implemented in various ways, such as CPU, GPU, CPU+GPU, ARM+FPGA, TPU+various SoCs, etc. Dr. Yao Yuan believes that FPGA chip structures are particularly suitable for model prediction. Using dynamic reconfiguration technology, they can quickly switch to different design schemes within a second to optimize hardware for the next workload. However, the development difficulty and barriers limit their widespread application. Vision Rui is dedicated to solving this problem with the launch of the industrial inspection EagleGo solution, allowing engineers to complete system operation in just a few hours using the Xilinx SDSoC development process with C or C++ language. Dr. Yao Yuan also introduced use cases, such as using SDSoc technology to achieve convolution kernel acceleration and object classification cameras.
Next-generation embedded development platform AWorks is the small cannon in RT Kernel
Zhou Ligong, Chairman of Guangzhou Zhiyuan Electronics Co., Ltd.
Zhou Ligong introduced that software design should aim to reduce overall costs. If human factors are considered, software engineering is closer to economics rather than computer science. Without changing the way of thinking, it is difficult to develop products that are both sellable and cost-effective. Therefore, Zhiyuan Electronics is also changing its way of thinking, using “chip + AWorks software platform” to design high-value-added modules, boards, and high-end measuring instruments, connecting to the ZWS IoT cloud through wired and wireless interfaces to achieve big data processing and form an industrial Internet ecosystem.
Zhou Ligong referred to AWorks as a small cannon in RT Kernel, possessing an extremely small native kernel that can run on platforms with a minimum of 1 KB RAM and 2 KB ROM; it can provide various OS services such as tasks, semaphores, mutexes, and message queues; it can support an unlimited number of tasks, with a maximum of 1024 task priorities, and supports tasks of the same priority; all components can be statically instantiated to avoid memory leak risks. AWorks components are designed with a pluggable and customizable concept, allowing for a tailored operating system. Additionally, the now very tricky USB driver can be implemented using the AWorks platform with just two lines of code to directly call stable USB devices, supporting all mainstream USB flash drives on the market.
AWorks is released in various forms, including wireless core versions, standard core boards, and SDK development packages, allowing engineers to experience a different embedded development experience.
AI and edge computing are extremely valuable in the IoT
Zou Jun, CTO of Shenzhen Rihai Communication Technology Co., Ltd.
Zou Jun has his own understanding of the Internet of Things, believing that the core of the IoT is scale. However, currently, the IoT industry chain is too long, and the connections between various links are still not smooth, leading to serious fragmentation of industry applications and the formation of information silos. Many companies are eager to layout in the IoT industry, but the overall scale is not large. Therefore, in the IoT industry, the “platform + module” capability is key to strategic layout. Zou Jun believes that AI and edge computing will play an important role in the IoT.
ADAS system has a bright future
Zhang Weiwei, General Manager of Shenzhen Zhongke Longzhi Automotive Co., Ltd.
Zhang Weiwei introduced that 2020 will begin the golden decade of autonomous driving and predicted the development trend of ADAS: from 2017 to 2019, the focus will be on safety warning applications, mainly using monocular cameras for driver behavior monitoring, forward collision warning, road departure warning, driver identity monitoring, and pedestrian collision warning; from 2018 to 2020, it will focus on auxiliary intervention applications, mainly using radar + camera for adaptive cruise control, automatic emergency braking, traffic congestion assistance, road keeping assistance, and automatic parking; from 2020 to 2025, it will focus on autonomous driving in specific environments, mainly using sensors + V2X applications for road planning, control strategies, and vehicle control. Liu Weiwei used NIO ES8 as an example to introduce the application of ADAS.
In the intelligent era, robots are indispensable
Zhang Guoping, CEO of Shenzhen Dazhu Robot Co., Ltd.
As we enter the intelligent era, robots are becoming increasingly common, leading many companies to design robots for their familiar fields. Zhang Guoping introduced the collaborative robots, AGV mobile robots, and STAR from Dazhu Robotics, and shared with the engineers the key technologies used in robot design, including integrated drive control joint modules, large and hollow torque motor design, servo driver design, absolute encoder technology, high-speed bus technology, electromagnetic brake technology, sensor technology, and HD reducer assembly technology. He also stated that the company has achieved localization of core components. Currently, Dazhu’s products are mainly applied in industries such as manufacturing, medical, semiconductor, and services.
After the meeting, participating engineers and experts engaged in heated discussions on technical and industrial issues of concern. Everyone generally believes that embedded and IoT engineers should seize the great opportunities presented by the era of artificial intelligence, combining their mastered technologies with the needs of an intelligent society, and play a good role as the technical core! University teachers, senior professionals in enterprise management and investment from Shenzhen and other regions also participated in the exchange, and the Beihang University Press Embedded Systems Branch brought new publications on IoT and embedded books to the meeting.
The Embedded Systems Association, initiated and operated by well-known scholars and industry figures in China, has been established and operated for 10 years. The association provides an environment for academic and industrial information exchange for experts, scholars, engineering technicians, market and technology media personnel in various disciplines of embedded systems in China. It is a public welfare scientific salon organization that enhances personal friendship, and the meeting does not charge fees for speakers and participants.
1.Teach you how to learn FPGA and easily realize engineering dreams
2.Foreign programmers seek help on what to do after 40, see what replies they get!
3.Oh my! As an electronic engineer, you are still using Baidu to find information?
4.To learn STM32, you must understand these five embedded operating systems!
5.How important is the RF filter? A single article will make you understand
6.Many people find FPGA difficult to learn because they haven’t summarized that digital electronics ultimately leads to analog electronics!
