Indoor Human Activity Detection System Based on Millimeter-Wave Radar

Abstract

Millimeter-wave radar has strong application value in industrial sensing and intelligent monitoring, while indoor personnel detection based on visual sensors may introduce privacy risks and can be affected by illumination and occlusion. On the basis of the original FMCW millimeter-wave radar scheme, this paper condenses the system design and experimental verification of an indoor personnel detection system using the IWR6843/IWR6843AOP radar platform. The study keeps the radar signal-processing chain, including point-cloud generation, static clutter cancellation, CA-CFAR detection, Capon direction-of-arrival estimation, DBSCAN clustering, and Kalman or extended Kalman tracking. A Python upper-computer platform integrates serial communication, radar parameter configuration, multithreaded data parsing, OpenCV visualization, Tkinter control, and Flask Web monitoring. Indoor tests show that the system can detect, count, locate, and track moving personnel in a 3 m×4 m room, and that target point-cloud density changes with range. The results also indicate that tracking accuracy decreases when multiple targets are close to each other or remain static, which provides a clear direction for improving filtering and tracking performance.