邊緣人工智慧晶片：智慧家庭和個人化人工智慧崛起的關鍵參與者

智慧家庭設備比以往任何時候都更加普遍，正在為我們的日常生活帶來廣泛的人工智慧 (AI) 功能，並開創一個注重大規模個性化的時代。 為了有效發揮作用，智慧家庭不僅依賴智慧家電，還依賴各種感測器和能源管理系統，這有助於家庭中物聯網設備的增加。 這些設備通常在微控制器單元 (MCU) 上使用 TinyML（微型機器學習），適用於功率和尺寸有限的消費性應用。這些設備中採用人工智慧晶片，可根據個人人工智慧需求提供快速運算。

在本報告中，我們分析了當前智慧家庭AI晶片的普及和利用，特別是個人化趨勢，並考慮了未來發展趨勢的展望。

目錄

第一章：大規模個人化時代：智慧家庭和個人化人工智慧需求不斷增長

• （1） Matter標準持續更新，加速智慧家庭進程
• （2）利用AI客製化個人化智慧家庭設備

第二章 廠商動態及發展趨勢

• （1） MCU與NPU融合，推動AI應用於智慧家居
• (2) RISC-V架構的興起提供了更多可客製化的產品選擇

第三章 TRI的觀點

• （一）智慧家庭將持續擁抱AI，Matter標準加速融合
• （2） RISC-V架構的興起與NPU整合MCU的普及

Smart home devices, now more common than ever, are bringing an array of AI capabilities into our daily routines, heralding the beginning of an era focused on large-scale personalization. To function effectively, smart homes rely not just on smart appliances but also on a variety of sensors and energy management systems, contributing to the growth of IoT devices in homes. These devices typically use Tiny Machine Learning (TinyML) on Microcontroller Units (MCUs), which is suited for their limited power and size in consumer applications. AI chips are employed within these devices to provide quick computing, tailored to individual AI needs.

Table of Contents

1. The Era of Large-Scale Personalization: Increased Demand for Smart Homes and Personalized AI

• (1) Continuous Updates to Matter Standards Driving Progress in Smart Homes
• (2) AI Customizes Personalized Smart Home Devices

2. Manufacturer Dynamics and Development Trends

• (1) Integrating MCUs with NPUs to Propel AI in Smart Homes
• (2) The Rise of RISC-V Architecture Offers More Customizable Product Choices

3. TRI's View

• (1) Smart Homes Continue to Embrace AI, with the Matter Protocol Accelerating Integration
• (2) The Rise of RISC-V Architecture and the Prevalence of NPU-Integrated MCUs