Multitarget Physical Activities Monitoring and Classification Using a V-Band FMCW Radar
Victor G. Rizzi Varela, Davi V. Q. Rodrigues, Leya Zeng, Changzhi Li
Abstract
This work proposes a radar-based physical activities monitoring solution capable of detecting, classifying, and counting the number of repetitions of multiple human subjects doing different exercises simultaneously in front of the radar, using only one monostatic radar. Four exercises were performed in the experiments: sit-ups, push-ups, squats, and jump rope which are very common exercises performed in and out of the gym without the need for expensive gym equipment. A 60-GHz frequency-modulated continuous-wave (FMCW) radar was used in the experiments. The extracted data from the radar were processed using MATLAB R2021b, and the features were extracted using range–Doppler frames and micro-Doppler (mD) analysis generating a spectrogram. By combining these two analyses, information about the range, time, Doppler, and radar cross section (RCS) can be extracted for each individual. To validate the feasibility and robustness of the work in realistic scenarios, static and moving clutter were added to some of the measurements.