Research Publications
Discover our latest research contributions to the scientific community. Our publications span journal articles, conference papers, and technical reports in radar technology and related fields.
Robust Multiobject Tracking Using mmWave Radar-Camera Sensor Fusion
A. Sengupta, L. Cheng, S. Cao
A robust tracking framework using high-level monocular-camera and mmWave radar sensor-fusion. Improves localization accuracy through decision-level sensor fusion and provides robustness via tri-Kalman filter setup for continuous tracking despite single sensor failures.
Stabilizing Skeletal Pose Estimation using mmWave Radar via Dynamic Model and Filtering
S. Hu, A. Sengupta, S. Cao
This work presents a method for stabilizing skeletal pose estimation using mmWave radar through dynamic modeling and filtering techniques. Addresses the challenge of unstable pose estimates in radar-based systems by incorporating temporal consistency constraints.
Automatic Radar-Camera Dataset Generation for Sensor-Fusion Applications
A. Sengupta, A. Yoshizawa, S. Cao
A novel approach that leverages YOLOv3 based highly accurate object detection from camera to automatically label point cloud data obtained from a co-calibrated radar sensor. Features co-calibration, clustering and association capabilities for automatically generating datasets containing labeled radar data-camera images for deep learning applications in sensor fusion.
mmFall: Fall Detection using 4D mmWave Radar and a Hybrid Variational RNN AutoEncoder
F. Jin, A. Sengupta, S. Cao
mmFall presents a hybrid variational RNN autoencoder architecture for fall detection using 4D mmWave radar. The system analyzes micro-Doppler and range-Doppler features with temporal modeling through RNN variants, achieving robust real-time fall detection while maintaining privacy. Achieves 98% detection rate out of 50 falls with only 2 false alarms.
NLP based Skeletal Pose Estimation using mmWave Radar Point-Cloud: A Simulation Approach
A. Sengupta, F. Jin, S. Cao
First approach to estimate 3D positions of 25 skeletal keypoints using simulated mmWave radar-like point-cloud data with natural language processing techniques. Demonstrates the potential of mmWave radars for sparse point-cloud representation with higher resolution than traditional radar counterparts, validated through simulation-based methodology.
mm-Pose: Real-Time Human Skeletal Posture Estimation using mmWave Radars and CNNs
A. Sengupta, F. Jin, R. Zhang, S. Cao
A real-time human skeletal pose estimation system using CNNs trained on mmWave radar data. The approach achieves efficient inference suitable for edge deployment while maintaining high accuracy in pose tracking applications. First method to detect more than 15 distinct skeletal joints using mmWave radar reflection signals.
MmWave Radar Point Cloud Segmentation using GMM in Multimodal Traffic Monitoring
F. Jin, A. Sengupta, S. Cao, Y. Wu
A multimodal traffic monitoring approach using high-resolution mmWave radar for point cloud representation. Applies multivariate Gaussian mixture model (GMM) for radar point cloud segmentation using point-wise classification in unsupervised learning for distinguishing pedestrians, cars, and bicycles. Demonstrates robust performance in adverse weather and lighting conditions.
Robust and Adaptive Radar Elliptical Density-Based Spatial Clustering and labeling for mmWave Radar Point Cloud Data
R. Zhang, S. Cao
A robust and adaptive radar point cloud clustering algorithm, named radar elliptical density-based spatial clustering of applications with noise (REDBSCAN). The proposed algorithm shows better clustering results for adapting to the arbitrary shape of targets as well as any number of targets comparing with traditional clustering methods, implemented using state-of-art mmWave radar sensor with MIMO antennas.
Extending Reliability of mmWave Radar Tracking and Detection via Fusion With Camera
R. Zhang, S. Cao
The proposed fusion system takes into consideration the error bounds of the two different coordinate systems from the heterogeneous sensors, and further a new fusion-extended Kalman filter is utilized to adapt to the heterogeneous sensors. Addresses real-world application considerations such as asynchronous sensors, multi-target tracking and association. Achieves range accuracy of 0.29m with angular accuracy of 0.013rad in real-time.
A DNN-LSTM based Target Tracking Approach using mmWave Radar and Camera Sensor Fusion
A. Sengupta, F. Jin, S. Cao
A sensor fusion study combining monocular camera and mmWave radar using deep neural networks and LSTMs. Presents a decision framework to produce reliable output when either sensor fails, leveraging radar's accurate depth information and camera's cross-range resolution. Experimental results demonstrate advantages over single-sensor approaches under uncertainty conditions.