AI-Powered Smart Refrigerator with User Profiling & Interaction Prediction

SmartCool is an edge-ready intelligent refrigeration system that transitions conventional refrigerators from reactive operation to context-aware, adaptive behavior. The system integrates mmWave radar, load cells, and an inertial measurement unit into a lightweight sensing architecture that captures user interactions and environmental dynamics. mmWave and accelerometer data are jointly used to infer user presence, motion characteristics, and interaction patterns, enabling behavior classification and implicit user identification through distinctive interaction signatures. The load cell subsystem monitors internal weight variations to detect and classify events such as restocking and item removal. All inference runs locally on embedded hardware using optimized models, preserving user privacy while maintaining low latency and low power consumption. Historical Beko usage datasets complement real-time sensing by providing longitudinal analysis of recurring usage patterns. The system is validated through real-time demonstrations and controlled experiments, achieving 93.9% interaction classification and 95.0% weight event detection accuracy, establishing a foundation for personalized, energy-efficient control in next-generation refrigeration systems.