The need for technologies for Human Activity Recognition (HAR) in home environments is becoming more and more urgent because of the aging population worldwide. Radar-based HAR is typically using micro-Doppler signatures as one of the main data representations, in conjunction with classification algorithms often inspired from deep learning methods. One of the limitations of this approach is the challenging classification of movements at unfavorable aspect angles (i.e., close to 90°) and of static postures in between continuous sequences of activities. To address this problem, a hierarchical processing and classification pipeline is proposed to fully exploit all the information available from millimeter-wave (mm-wave) 4D imaging radars, specifically the azimuth and elevation information in conjunction to the more conventional range, Doppler, received power, and time features. The proposed pipeline uses the two complementary data representations of Point Cloud (PC) and spectrogram, and its performance is validated using an experimental dataset with 6 activities performed by 8 participants. The results show good performance of the proposed pipeline compared with alternative baseline approaches in the literature, and the effect of key parameters such as the amount of training data, signal-to-noise levels, and virtual aperture size is investigated. Leave-one-subject-out test is also applied to study the impact of body characteristics on the generalizability of the trained classifiers.

@en

The feasibility of classifying human activities measured by a distributed ultra-wideband (UWB) radar system using Range-Doppler (RD) images as the input to classifiers is investigated. Kinematic characteristics of different human activities are expected to be captured in high-resolution range-Doppler images measured by UWB radars. To construct the dataset, 5 distributed monostatic Humatics P410 radars are used to record 15 participants performing 9 activities in arbitrary directions along a designated trajectory. For the first time a convolution-free neural network based on the novel multi-head attention mechanism (the Vision Transformer architecture) is adopted as the classifier, attaining an accuracy of 76.5 %. A comparison between Vision Transformer and more conventional CNN-based architectures, such as ResNet and AlexNet, is also conducted. The robustness of Vision Transformer and the other networks against unseen participants is also validated by testing via Leave One Participant Out validation.@en