Multiple-person Tracking with mmWave Radar Using Adaptive Clustering and Ghost Removing

This paper describes a new system for tracking multiple people indoors using mmWave radar. The proposed system addresses three main problems: clustering failure, ghost targets, and detection of stationary individuals. This research proposes a novel adaptive clustering algorithm that improves the accuracy of tracking multiple individuals. Moreover, an algorithm that targets and removes ghost artefacts, i.e., false detections caused by reflections from walls and other objects, is introduced. Finally, a Doppler-based algorithm is used to detect stationary individuals. This is particularly useful in care facilities where elderly people may spend long periods of time sitting or lying down. The proposed system was tested in various scenarios and was found to be able to track individuals or pairs of elderly people highly accurately. However, when tracking three or more elderly people, the accuracy rate dropped significantly. It is concluded that further improvements are needed to achieve higher accuracy when tracking multiple individuals in places like nursing homes.