Abstract:This study explored an indoor system for tracking multiple humans and detecting falls, employing three Millimeter-Wave radars from Texas Instruments. Compared to wearables and camera methods, Millimeter-Wave radar is not plagued by mobility inconveniences, lighting conditions, or privacy issues. We conducted an initial evaluation of radar characteristics, covering aspects such as interference between radars and coverage area. Then, we established a real-time framework to integrate signals received from these radars, allowing us to track the position and body status of human targets non-intrusively. Additionally, we introduced innovative strategies, including dynamic Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering based on signal SNR levels, a probability matrix for enhanced target tracking, target status prediction for fall detection, and a feedback loop for noise reduction. We conducted an extensive evaluation using over 300 min of data, which equated to approximately 360,000 frames. Our prototype system exhibited a remarkable performance, achieving a precision of 98.9% for tracking a single target and 96.5% and 94.0% for tracking two and three targets in human-tracking scenarios, respectively. Moreover, in the field of human fall detection, the system demonstrates a high accuracy rate of 96.3%, underscoring its effectiveness in distinguishing falls from other statuses.
engineering, electrical & electronic,instruments & instrumentation,chemistry, analytical
What problem does this paper attempt to address?
### Problems Addressed by the Paper
This paper aims to address the issues of multi-person tracking and fall detection in indoor environments. Specifically, the authors utilize three millimeter-wave radars from Texas Instruments to develop a real-time system for non-intrusive human position tracking and state recognition.
#### Main Contributions:
1. **Multi-Radar Deployment**: By deploying three radars to expand the coverage area, a real-time system was designed to capture point cloud data from the scene at a rate of 20 frames per second.
2. **Innovative Strategies**: Introduced a dynamic density-based spatial clustering algorithm (DBSCAN) to enhance target detection, a probability matrix for multi-person tracking, and target state prediction for fall detection.
3. **Performance Evaluation**: Through over 300 minutes of data experiments (approximately 360,000 frames), the tracking accuracy in single-person, two-person, and three-person scenarios was demonstrated to be 98.9%, 96.5%, and 94.0%, respectively, with a high fall detection accuracy of 96.3%.
4. **Open Source Code**: The research work was open-sourced to further advance research in this field.
### Research Background and Related Work
- **Millimeter-Wave Radar Technology**: Millimeter-wave radars operate in the high-frequency range (76 to 81 GHz), offering high resolution and anti-interference capabilities, making them suitable for various indoor human activity recognition (HAR) applications.
- **Comparison with Existing Methods**: Compared to wearable devices (such as inertial measurement units, accelerometers, etc.) and camera-based methods, millimeter-wave radars offer higher privacy protection and adaptability to lighting conditions.
### Experimental Setup and Data Collection
- **Radar Configuration**: Three Texas Instruments IWR1843 millimeter-wave radars were used, installed on the wall, at a 55-degree tilt angle, and on the ceiling to cover more human reflection signals.
- **Data Collection**: Data from 15 randomly selected participants were collected, totaling approximately 300 minutes (about 360,000 frames), covering walking, sitting, standing, and falling actions in single-person, two-person, and three-person scenarios.
### System Design
- **Modular Architecture**: The system was developed using Python 3.8 and includes modules for data reading, early processing, post-processing, visualization, and queue monitoring.
- **Data Preprocessing**: Rotation and position compensation were performed on multi-radar data to ensure correct alignment in a unified coordinate system.
- **Background Noise Elimination**: Background noise was reduced using a global filter (Global BES_Filter) and dynamic DBSCAN algorithm to improve data quality.
- **Target Detection and Tracking**: Dynamic DBSCAN was used for human target detection, combined with a probability matrix to achieve multi-person tracking and fall detection.
Through these methods, the system achieved efficient and accurate real-time multi-person tracking and fall detection, performing excellently in various application scenarios.