Raspberry Pi 4B-driven smart home system with multimodal fusion

This study develops a smart home system based on Raspberry Pi 4B to realise home intelligent control through an efficient and economical hardware platform. The system integrates face recognition, voice interaction, anti-theft alarm, device control and environment monitoring functions, using temperature and humidity, smoke and infrared sensors to collect data and upload to the MQTT server. Users can view the home environment in real time and control devices remotely through WeChat applets. Adopting cloud-side collaborative computing architecture, the system achieves 62.8% accuracy in facial recognition, 100% success rate in voice recognition, and response time below 200 ms. Experiments prove that the system has superior stability, practical value and cost-effectiveness.