RadarTrack: Enhancing Ego-Vehicle Speed Estimation with Single-chip mmWave Radar
RadarTrack is a lightweight, real-time ego-speed estimation framework using a single mmWave radar and a novel phase-based method, enabling robust performance on embedded platforms without relying on deep learning or static environments.
Key Features
- Analytic Phase Equation: Fourth-order kinematics-based equation for phase change enables accurate ego-speed estimation.
- Cross-Platform Radar Solution: Lightweight ego-motion estimation using a single COTS mmWave radar on UGVs, drones, and handheld devices.
- Dynamic & Static Segmentation: Doppler-based speed profiling effectively separates moving and stationary objects in complex scenes.
- Real-Time Edge Processing: Achieves ≈0.29s latency on Jetson Nano without any deep learning, 85% faster than SOTA (Radarize).
Licensing:The platform source and the collected dataset is free to download and can be used with GNU Affero General Public License for non-commercial purposes. The institute's ethical review committee has approved the field study (Order No: IIT/SRIC/DEAN/2023, Dated July 31, 2023).
Contributors
Argha Sen
IIT Kharagpur, India
Soham Chakraborty
IIT Kharagpur, India
Soham Tripathy
IIT Kharagpur, India
Sandip Chakraborty
IIT Kharagpur, India
Teaser Video
Publications
- Argha Sen, Soham Chakraborty, Soham Tripathy, and Sandip Chakraborty. "Poster: Dynamic Ego-Velocity Estimation Using Moving mmWave Radar: A Phase-Based Approach." ACM MobiSys Posters, 2024.
- Argha Sen, Soham Chakraborty, Soham Tripathy, and Sandip Chakraborty. "RadarTrack: Enhancing Ego-Vehicle Speed Estimation with Single-chip mmWave Radar." IEEE SmartComp 2025
- Argha Sen, Soham Chakraborty, Soham Tripathy, and Sandip Chakraborty. "DEMO: Beyond Doppler -- Demonstrating Phase-Based Ego-Speed Estimation on Embedded mmWave Radar." IEEE SmartComp Demos, 2025
