‘엣지 디바이스 및 온 디바이스 AI를 위한 레이더 센싱 기술’ 웨비나에 앞서 발표자인 최재영 인피니언 코리아 전력&센서 시스템 사업부 매니저의 이야기를 들어보았다.
“Designing a sensing system for edge AI utilization, prioritizing user analysis”
Trends in activating product operation through real-time sensing and enhancing existing sensor functions through deep learning
Importance of radar's biosignal/behavioral pattern analysis/gesture recognition combined detection function ↑
Infineon to Hold Webinar on Radar Sensing Technology for Edge and On-Device AI on July 16
[Editor's Note] Recently, as smart sensor technology and AI application fields are expanding, the need for AI sensing technology is emerging. The requirement for performing complex functions is increasing through activating product operation through real-time sensing in edge devices and strengthening existing sensor functions and adding new functions through deep learning or machine learning in on-device. In particular, in the case of radar, the importance of complex sensing functions such as biosignal analysis, behavioral pattern analysis, or gesture recognition is also increasing. Through a webinar with Infineon Korea on July 16, our magazine will examine the configuration and operation of a radar system capable of complex sensing functions in edge devices and on-devices, introduce Infineon radar products, and share utilization plans for each industry and application. Ahead of the webinar on 'Radar Sensing Technology for Edge Devices and On-Device AI', we heard from the presenter, Jaeyoung Choi, Manager of the Power & Sensor Systems Business Unit at Infineon Korea.
.jpg)
▲Choi Jae-young, Manager of Power & Sensor Systems Business Unit, Infineon Korea
■ Please explain the differences between existing sensing technology and sensing for edge devices and on-device AI.
The biggest difference and core is that it is a developer-centric sensor system and a user-centric sensor system.
It can be said that the system design purpose and usage are slightly different.
Existing radar sensors are manufactured and designed from an engineering perspective, with a focus on sensing performance.
On the other hand, as sensing is performed to utilize AI in edge devices and on-devices, the purpose of system design is changing from considering only improving detection performance to prioritizing various analyses on users.
In other words, the biggest difference is that it has expanded from a sensor system that only considered engineering thinking to a user-centered system with added humanistic concerns.
■ What are the advantages of applying radar sensors to edge devices and on-devices?
Edge devices and onRadar sensing for utilizing AI in devices can maximize data utilization and efficiency by utilizing separate AI engines or learned AI algorithms for additional functions in addition to existing simple function processing.
Existing radars are based on estimating information about the distance, speed, and angle of a target, and aim to improve the accuracy of that information and to apply the acquired information in one dimension, such as tracking or presence detection.
And for this purpose, signal processing was performed through explicit programming.
AI-type radar sensing was possible through the cloud or a separate computing server, but it was difficult to expect a large practical effect due to physical and environmental limitations.
On the other hand, radar sensing utilizing AI in edge devices or on-devices can increase both data usability and efficiency, as well as response speed and expanded functionality through efficient processing and management of data.
In particular, it is possible to independently perform signal processing within the sensor module itself or within the product equipped with the sensor, and to additionally flexibly utilize AI to analyze user behavior, situations, or specific environments in addition to the existing general functions.
This has the advantage of being able to efficiently expand and deploy the role or processing scope of the sensor itself compared to existing radar sensors or AI-based radar sensing using the cloud.
■ What are the features of Infineon’s radar sensor?
Infineon's radar sensors have already been proven in the market for their performance, price, and product reliability.
Especially Infineon's Xensiv 60GHz Radar offers a high degree of freedom in radar design for edge devices or on-device AI utilization.
In addition, radar sensors for various purposes can be utilized with different differentiation points through the application of unique signal processing configurations or new types of algorithms rather than the uniform signal processing structure and limited algorithm configurations.
This can become a powerful smart sensor solution that can expand the use of human detection sensors and various functions in products and provide differentiated value to products.
■ It is thought that the utilization method will be different for each industry and application. Could you please tell me what is different and whether Infineon's radar sensor can handle all of this?
To meet the requirements and conditions of various fields and applications, it is necessary to have a variety of solution products.
Infineon's Xensiv radar family offers a variety of solutions to address both Doppler CW radar and FMCW radar.
We are also continuously developing products ranging from 1Tx1Rx to 2Tx4Rx to enable the use of radars with different characteristics through various channel configurations as a single product.
This family of radars is already being used in a variety of applications in portable devices, TVs, air conditioners, home appliances and automobiles.
It is being used in many product groups and for medical purposes, from simple presence detection functions to monitoring functions for obtaining data on the user's location and movement, and healthcare functions through analysis of behavior, situation, and biosignals.
In addition, it can be used as a risk detection function for industrial safety management or as an entertainment function for user convenience.
■ What is the main content of the July 16 webinar?
This webinar will discuss ways to effectively utilize radar sensors in the AI era.
Radar sensors have been used in various fields and applications for a long time.
Radar sensors are no longer novel, new, difficult technologies or tricky to apply, but rather are common sensor products.
So the popular radar sensorThis paper aims to explain the design and system concept perspectives of how to configure a system and implement complex functions for utilizing AI in edge devices or on-devices.
■ Please say a word to e4ds readers
In the midst of recent technological trends such as AI, green energy, and digitalization, the presence of differentiated value is becoming increasingly important.
We have prepared this to help you understand and gain knowledge about how to easily and effectively utilize AI technology through radar.
We hope many of you will attend.