Fall Detection and Assessment Using Multitask Learning and Micro-sized LiDAR in Elderly Care

The increasing concern over the rapid aging population has brought to light a significant issue: the escalating number of falls among the elderly. As seniors grow older, they become more susceptible to physical ailments, leading to a higher frequency of falls. The lack of prompt assistance after a fall further compounds the problem, putting them at risk of severe consequences, including mortality. To address this pressing matter, the demand for fall detection systems in nursing homes and similar care facilities is on the rise. In response, our research proposes developing a point cloud-based fall detection system, complete with risk assessment capabilities, catering to the specific needs of the elderly. By employing advanced 3D LiDAR technology, we can scan the environment while preserving privacy. The algorithm employed then carefully analyzes this representation, extracting spatio-temporal discriminative features, thus enabling accurate fall detection. We have thoroughly evaluated the proposed system using collected data in our lab, and the results are promising, demonstrating its ability to detect fall events effectively. The successful implementation of this system could significantly enhance safety and improve the overall quality of life for the elderly population.