Lightweight Range-Angle Imaging Based Algorithm for Quasi-Static Human Detection on Low-Cost FMCW Radar

TL;DR

The proposed lightweight, non-visual image-based method for robust quasi-static human presence detection using a low-cost 60 GHz FMCW radar demonstrates that simple image-based processing can provide robust and deployable quasi-static human sensing in cluttered indoor environments.

Abstract

Quasi-static human activities such as lying, standing or sitting produce very low Doppler shifts and highly spread radar signatures, making them difficult to detect with conventional constant-false-alarm rate (CFAR) detectors tuned for point targets. Moreover, privacy concerns and low lighting conditions limit the use of cameras in long-term care (LTC) facilities. This paper proposes a lightweight, non-visual image-based method for robust quasi-static human presence detection using a low-cost 60 GHz FMCW radar. On a dataset covering five semi-static activities, the proposed method improves average detection accuracy from 68.3% for Cell-Averaging CFAR (CA-CFAR) and 78.8% for Order-Statistics CFAR (OS-CFAR) to 93.24% for Subject 1, from 51.3%, 68.3% to 92.3% for Subject 2, and 57.72%, 69.94% to 94.82% for Subject 3, respectively. Finally, we benchmarked all three detectors across all activities on a Raspberry Pi 4B using a shared Range-Angle (RA) preprocessing pipeline. The proposed algorithm obtains an average 8.2 ms per frame, resulting in over 120 frames per second (FPS) and a 74 times speed-up over OS-CFAR. These results demonstrate that simple image-based processing can provide robust and deployable quasi-static human sensing in cluttered indoor environments.

Figures (6)

• Figure 1: End-to-end radar's imaging processing pipeline. Raw complex IQ frames from the 60 GHz FMCW radar are converted to range–Doppler maps via range and Doppler FFTs with DC removal, windowing, and MTI clutter suppression. Capon beamforming over two receive channels produces a normalized range–azimuth (RA) intensity map, which is then processed by three detectors: 2-D CA–CFAR, 2-D OS–CFAR, and the proposed percentile-gated lump detector.
• Figure 2: Capon-based range–azimuth surface response (1 frame) for a subject lying on the sofa. The subject produces a compact high-intensity peak in range–azimuth space on top of a broad background clutter ridge from walls and furniture.
• Figure 3: Percentile ablation on a representative range--azimuth frame (lying on sofa). From left to right: normalized RA image and detection masks obtained by keeping only the top 90%, 95%, and 99% of pixel intensities.
• Figure 4: Qualitative comparison of all three detectors on frame 21 for subject 1 lying on the floor, which all of them produce detections within the ground-truth bounding box.
• Figure 5: Qualitative comparison of all three detectors on frame 106 for subject 1 lying on the floor. only our proposed method can detect within the ground-truth bounding box.