Activity Sensing System for Elderly People Living Alone Based on Millimeter Wave Radar and Depth Camera

With the global population aging, the need to ensure the health and safety of elderly individuals living alone has become a pressing societal issue. This paper introduces an activity sensing system utilizing millimeter-wave radar and depth camera technology to monitor the movements and behaviors of elderly people residing independently in home environments. The system's architecture features precise indoor positioning and trajectory tracking through the integration of data from both the millimeter wave radar and the depth camera, enhanced by cross-modal supervision. Employing a neural network, the system learns the relationships between the radar and camera data, which significantly improves positioning accuracy. Additionally, a lightweight convolutional neural network is integrated to detect hazardous situations, such as falls, facilitating swift emergency responses. Extensive experiments validate the system's performance, demonstrating successful tracking of movement trajectories and accurate recognition of activities including walking, standing, falling, and sitting. The system also includes a user-friendly front-end platform that facilitates real-time observation of the elderly's location and activities, fostering remote monitoring capabilities. The proposed activity sensing system offers a privacy-preserving and environmentally robust solution tailored to address the growing challenges of elderly care and independent living scenarios.