Realistic Micro-Doppler Radar Simulation of Cyclists for Vulnerable Road User Classification

Sura O, Mergenthaler P, Kammel C, Dorschky E, Hoffmann M, Vossiek M (2025)

Publication Type: Conference contribution

Publication year: 2025

Publisher: IEEE

City/Town: New York City

Pages Range: 315-318

Conference Proceedings Title: 2025 22nd European Radar Conference (EuRAD)

Event location: Utrecht

DOI: 10.23919/EuRAD65285.2025.11233995

Abstract

Accurate classification of Vulnerable Road Users (VRUs) in diverse scenarios is crucial for automotive radar applications. However, obtaining suitable training data, especially for edge cases and accident scenarios, presents a significant challenge. This paper introduces a novel approach for realistically simulating cyclists, capturing detailed micro-Doppler components and motion features from both the bicycle and the cyclist. An optical motion-capture system is used to create a digital twin of a cyclist, which is then validated through comparison with simultaneous real-world measurements. This validation ensures the accuracy of the simulation model. The resulting simulation data, featuring automatically generated labels, can be used to efficiently train AI systems and apply them in real-world scenarios. This approach enables precise and repeatable modeling of edge cases and complex pre-crash scenarios, supporting the reliable deployment of AI for VRU classification tasks.

Authors with CRIS profile

How to cite

APA:

Sura, O., Mergenthaler, P., Kammel, C., Dorschky, E., Hoffmann, M., & Vossiek, M. (2025). Realistic Micro-Doppler Radar Simulation of Cyclists for Vulnerable Road User Classification. In 2025 22nd European Radar Conference (EuRAD) (pp. 315-318). Utrecht, NL: New York City: IEEE.

MLA:

Sura, Oliver, et al. "Realistic Micro-Doppler Radar Simulation of Cyclists for Vulnerable Road User Classification." Proceedings of the 2025 22nd European Radar Conference (EuRAD), Utrecht New York City: IEEE, 2025. 315-318.

BibTeX: Download