Why do MCU manufacturers accelerate the deployment of edge AI?

Smart homes, industrial Internet of Things, wearable devices, and smart surveillance are currently hotspots that people are paying attention to. Improving the AI processing capabilities of edge computing will greatly increase the level of intelligence of terminal equipment. In this case, edge AI has entered the fast lane of development in recent years. The MCU is the most widely used processor in the field of edge computing, and its integration with edge AI has also become an important development trend.

MCU manufacturers accelerate the layout of edge AI

In recent years, the development of edge AI has become faster and faster. Due to the rapid increase in the amount of data generated by terminal devices, in order to achieve rapid response and parallel big data processing in the cloud, more and more manufacturers are turning their attention to edge computing, trying to embed trained AI algorithms in terminal devices, thereby greatly improving Computing power on the terminal side. However, in order to realize this idea, the edge AI processor should have low cost and low power consumption characteristics while possessing certain computing efficiency. MCU is undoubtedly one of the choices that can achieve the best results, so more and more MCU manufacturers are beginning to integrate MCU and edge AI.

Renesas has launched an embedded AI technology "e-AI" for this purpose. For example, in the field of intelligent manufacturing, the "e-AI" unit solution can be added to the device as an additional unit. Through the pre-learned AI processing model, the entire range from sensor data collection to data processing, analysis and evaluation/judgment can be realized. process. STMicroelectronics slogan "Let most STM32 products support AI deep learning" at the 4th STM32 Summit held in Shenzhen last year. Silicon Labs also cooperated with the artificial intelligence innovation company Cartesiam to optimize algorithms for the Internet of Things. In the demonstration watched by the reporter, it was shown that edge AI can be applied in the industrial Internet of things such as motor control. For example, a machine learning algorithm is used to train the fan operating mode. When it detects an abnormality in the device, it will send an alarm to the display via Bluetooth.

Domestic MCU manufacturers also recognize this development trend. It is reported that Qianxin Technology is working with Zhaoyi Innovation to carry out artificial intelligence ecological cooperation in the field of artificial intelligence and RISC-V ecology, and jointly provide customers with AI algorithm chip-level acceleration solutions based on TensorChip and Zhaoyi Innovation GD32 MCU technology. Through algorithm compression and computing core collaboration technology, the advantages of GD32V series MCU products in the field of AIoT are further strengthened, and customers in the corresponding fields are provided with more and more convenient integrated solution options.

Application landing, massive terminal intelligence emerges

"MCU+Edge AI" has begun to be applied in more and more fields. "Edge computing refers to sinking computing and processing capabilities from cloud data centers to access networks very close to users. For MCU manufacturers, in addition to the need to continuously optimize the integration, power consumption, cost, and security of the chip itself. In addition to building a broad product platform, it also needs to be advanced from multiple dimensions to meet the endless application needs." Jin Guangyi, Director of Zhaoyi Innovation Product Marketing, pointed out. Incorporating some simple artificial intelligence algorithms into the MCU is a supplement and enhancement to the existing MCU product series. With the advent of 5G, people will put forward higher and more extreme requirements for the delay and energy consumption of traditional products, and edge computing is a very good way to achieve it. The significance of the realization of artificial intelligence on the MCU is that it can combine the characteristics of low power consumption, low cost, real-time performance, stability, short development cycle, and broad market coverage of the MCU with the powerful processing capabilities of artificial intelligence, thereby enabling massive Terminal intelligence emerged.

Image and voice processing is one of the target application areas of "MCU+Edge AI". For example, graphics recognition, voice assistant wake-up word processing, and other applications such as voice classification for various security systems. In the field of intelligent manufacturing, due to the impact of production in many factories during the epidemic, coupled with the increase in labor costs, the trend of intelligent and mechanized industrial manufacturing will become deeper. In addition, the popularization of 5G will promote the development of Internet of Things, Internet of Vehicles and Wi-Fi6. At the same time, the wider coverage of 5G base stations will also bring about a wave of "replacement", which will give rise to more consumer electronics needs around 5G ecology, smart homes, and the Internet of Things. Therefore, the demand for MCUs will also grow rapidly in the next few years. . Chen Guodong, senior marketing manager for China of Cypress's Internet of Things Computing and Wireless Division, said: "In recent years, the country has been vigorously supporting 5G, electric vehicles, high-end manufacturing, consumer electronics, Internet of Things and other fields. These directions will become the future. The engine of sustained economic development is also a market that the MCU industry must pay attention to in the next few years. Under these general directions, some sub-sectors will emerge, such as smart door locks and smart speakers."

In the post-epidemic era, the market for wearable devices has also expanded. “Recently, NBA teams have begun to use smart rings to detect basic physical parameters of players to warn of the new crown virus. This smart ring is based on Cypress’s PSoC series of MCUs. After the epidemic, people’s attention to personal health will increase. So as to promote the development of the wearable device market. Existing devices are mainly for routine detection of heart rate and blood pressure. In the future, more functions will be covered, including back-end big data analysis, mining, and disease association research, etc. , These also need to be applied to edge AI technology." Chen Guodong said.

The balance of power consumption and performance is a challenge

If you want to realize AI applications in all areas of life in the future, MCU is undoubtedly one of the best choices. However, the performance of the MCU is relatively low, so there are still certain challenges in applying the MCU to AI operations.

"With the rapid development and popularization of cloud computing, edge computing has begun to gain more and more attention in the field of artificial intelligence. There are many reasons for this phenomenon. For example, huge data streams cause heavy network burdens, security issues, and system Latency, etc. In order to improve the overall performance of MCUs used for edge computing, the industry expects MCUs to have high processing power, ultra-low power consumption, ultra-small size, enhanced security mechanisms, etc." Silicon Labs, Asia Pacific Marketing Manager, IoT Products Qiu Yi pointed out.

How to achieve a balance between power consumption and performance is a problem that manufacturers need to consider. "At present, the development of the artificial intelligence industry requires the implementation of scenarios, and the requirements of different scenarios and applications are very different. It is far from enough to achieve voice recognition and face recognition. The core competitiveness of the subsequent industry is high integration, Ultra-low power consumption and high cost performance. Taking smart door locks as an example, while realizing face recognition, can the processor also integrate fingerprint recognition, door lock control and other system functions? Artificial intelligence must be applied to the scene, and then The functions are improved for this scenario. Therefore, it is necessary to improve the MCU's integration based on this scenario, while reducing power consumption and improving cost performance." Chen Guodong said.

In addition, the manufacturing process is an important consideration when MCU manufacturers carry out product design. "The next generation of MCU technology will develop to 40 nanometers and 22 nanometers. 40 nanometers and 22 nanometers are not cutting-edge technology nodes, but they are sufficient for MCUs. In these two sizes, MCUs can achieve the best cost. The more advanced the nodes, the MCU The better the dynamic power consumption, the worse the static power consumption, so you must find a balance. 40nm and 22nm are very suitable for MCU technology nodes." Chen Guodong pointed out.

Commax-Tech Electronic Co., Ltd      Electronic component specialist

B23, second floor, ASEAN building, Longhua District, Shenzhen

sales@commax-tech.com

https://commax-tech.com

Keyword：MCU manufacturers   Accelerate layout edge AI   Smart home   Industrial Internet of Things   Wearable devices and smart monitoring   5G   Electric vehicles   High-end manufacturing   Consumer electronics   Internet of things   Smart door locks   Smart speakers   Commax-Tech Electronic