Real-Time Smart Home Monitoring and Control Using Raspberry Pi, Sensors, and Edge AI

This paper presents the design, implementation, and performance evaluation of AI-Enhanced Pi-Home, an edge-AI-based smart home automation system integrating multimodal input recognition, environmental sensing, and real-time actuator control. The architecture is structured around a Raspberry Pi 4 core with layered modules for data acquisition, embedded analytics, and user interaction. Key components include DHT22, MQ-2, PIR, ESP32-CAM, ultrasonic sensors, and ACS712 current sensors, managed via Python scripts and interfaced using GPIO, I2C, and UART protocols. Recognition modules encompass facial detection, voice command interpretation, and gesture classification using pretrained embeddings and lightweight CNNs. Real-time monitoring, energy estimation, and automation scheduling are implemented through Flask-based dashboards and SQLite logs. Experimental results demonstrate system responsiveness under 1.5 seconds, automation accuracy over 92%, and energy savings up to 40% across test scenarios. Recognition accuracies for voice, gesture, and facial input achieved 93%, 89%, and 96% respectively. The platform supports real-time alerts, multimodal access control, and contactless interaction, making it viable for secure and energy-efficient smart residential applications.