RFiDARFusion: Enhancing Contactless Activity Monitoring with Radar and RFID Fusion

Indoor human activity recognition (HAR) faces challenges due to the limitations of single-sensor systems in terms of privacy and cost. Our contactless RFiDARFusion system aims to reduce healthcare costs by swiftly identifying unforeseen activities through artificial intelligence and wireless communication. The system accurately discerns complex human activities by employing the LSTM-VAE fusion algorithm for multi-sensor data extraction, proving effective in long-range and non-line-of-sight environments. Two data fusion algorithms were studied across five distinct activities, significantly improving HAR accuracy by a minimum of 4.5% and 6.5% under data and feature-level fusion, respectively. The resulting RFiDARFusion system is contactless, precise, and robust, promising to enhance the quality of life for elderly patients in assisted living.