A multi-target detection and position tracking algorithm based on mmWave-FMCW radar data

• DOI: 10.1016/j.measurement.2024.114797
• Publication Date: 2024-05-03T01:28:46Z
• Authors: Farhad Shamsfakhr; David Macii; Luigi Palopoli; Michele Corrà; Alessandro Ferrari; Daniele Fontanelli
• Keywords: Position (finance); Tracking (education)

Detecting and tracking the position of multiple targets indoors is a challenging measurement problem due to inherent difficulty cluster correctly sensor data associated given target track each with adequate accuracy. This critical especially in rooms filled fixed or moving objects hampering detection whenever paths different cross one another. In this paper, robust Multiple Targets Tracking (MTT) algorithm exploiting clouds points collected from mmWave-FMCW radar presented. The proposed solution consists four main steps. First, possible outliers raw set are removed using neural network model. Next, cleaned-up clustered Density-Based Spatial Clustering Applications Noise (DBSCAN) algorithm. Then, Kalman Filter (KF) used centroid cluster. Finally, Structured Branching Hypothesis Testing (SBMHT) applied updated over reasonably short time intervals decide which detected tracks supposed be confirmed ones instead should discarded. MTT technique was validated experimentally sets 60-GHz TI IWR6843 platform. reported results show that developed algorithm, if properly tuned, faster returns more accurate than other techniques. particular, percentage errors negligible planar positioning accuracy within about 30 cm 90% probability when up five move freely same room.