A5073606569- Advanced SAR Imaging Techniques
- Robotics and Sensor-Based Localization
- Radar Systems and Signal Processing
- Non-Invasive Vital Sign Monitoring
- Gait Recognition and Analysis
- Gamma-ray bursts and supernovae
- Optical measurement and interference techniques
- Space Satellite Systems and Control
- Indoor and Outdoor Localization Technologies
- Target Tracking and Data Fusion in Sensor Networks
- CCD and CMOS Imaging Sensors
Delft University of Technology
2022-2025
Unconstrained human activities recognition with a radar network is considered. A hybrid classifier combining both CNNs and RNNs for spatial-temporal pattern extraction proposed. The two-dimensional (2D-CNNs) are first applied to the data perform spatial feature on input spectrograms. Subsequently, gated recurrent units bidirectional implementations used capture long- short-term temporal dependencies in maps generated by 2D-CNNs. Three NN-based fusion methods were explored compared utilize...
The problem of instantaneous ego-motion estimation with mm-wave automotive radar is studied. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DeepEgo</i> , a deep learning-based method, proposed for achieving robust and accurate estimation. A hybrid approach that uses neural networks to extract complex features from input point clouds applies weighted least squares (WLS) motion utilized in . Additionally, novel loss function,...
The problem of estimating the 3D ego-motion velocity using multi-channel FMCW radar sensors has been studied. For first time, estimation is treated raw signals. A robust algorithm to instantly determine complete motion state ego-vehicle (i.e., translational speed and rotational speed) proposed. angle information targets extracted, then their phase from different times instances used vehicle through an optimization process. Any pre-processing steps, such as clustering or clutter suppression,...
Linked to the increasing availability of datasets for radar-based human activity recognition (HAR), in this Student Highlights contribution, we report on a classification project that group 23 graduate students performed at TU Delft. The were asked work groups 2–3 members and use publicly available University Glasgow dataset develop best pipeline as possible. This involved development justification both choices preprocessing techniques radar data (e.g., time–frequency distributions cleaning...
The problem of 2D instantaneous ego-motion estimation for vehicles equipped with automotive radars is studied. To leverage multi-dimensional radar point clouds and exploit features automatically, without human engineering, a novel approach proposed that transforms into weighted least squares (wLSQ) using neural networks. Comparison existing methods done challenging real-world dataset. comparison results show the method can achieve better performance in terms accuracy, long-term stability,...