EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Physics-Guided Multi-GSO Spectral Filtering for Degradation-Aware Automotive Radar Point-Cloud Detection

Xiuping Li iD, Xiyan Sun iD, Yuanfa Ji iD, Jingjing Li iD, Wentao Fu iD, Songke Zhao iD, Wenbin Liang iD, Xizi Jia iD, Jian Liu iD

Sensors · 24.07.2026 · 26 (15) · 4714

DOI: 10.3390/s26154714 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 24.07.2026. Verfasst von Xiuping Li und 8 weiteren Autor:innen. Beteiligte Einrichtung: School of Information and Communication, Guilin University of Electronic Technology, Guilin 541004, China. Die Arbeit nennt 53 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Guangxi Science and Technology Department, National Natural Science Foundation of China, Nanning Scientific Research and Technology Development Program.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automotive millimeter-wave radar produces sparse point clouds with Doppler velocity and radar cross-section (RCS), but graph detectors typically use a shared representation for semantic prediction and box regression despite their different propagation requirements. We propose multi-GSO spectral filtering (MGSF), a residual module that filters radar features over geometry-, Doppler-, and RCS-defined graph shift operators and fuses diffusion and residual components with a node-adaptive gate. MGSF-TD applies full multi-GSO refinement to semantic prediction and geometry-only refinement to box regression. On the complete RadarScenes validation set, MGSF-TD improves the official RadarGNN checkpoint from 60.19 to 60.59 mAP and from 74.06 to 75.10 mean foreground F1 (FG-F1). Across three MGSF-TD training seeds, the FG-F1 margin under RCS noise increases from +1.15 at 3 dBsm to +2.38 at 20 dBsm; seed-42 full-validation mAP margins are +0.24, +1.07, and +1.82. Controls show that geometry-only diffusion explains part of the gain and the RCS operator contributes most clearly at low-to-moderate noise, whereas a parameter-matched widened baseline matches or exceeds MGSF-TD under severe RCS and Doppler corruption. Cross-sensor diagnostics reproduce the Doppler failure trend but not the severity-dependent RCS gain. MGSF-TD therefore offers a balanced, physically interpretable operating point rather than a universal robustness gain.

Schlagwörter

Stichwörter der Autor:innen

Doppler velocityautomotive radarautonomous drivinggraph neural networkgraph signal processingintelligent transportation systemsobject detectionradar point cloudsensor degradationspectral filtering

Autor:innen und Institutionen

  • School of Information and Communication, Guilin University of Electronic Technology, Guilin 541004, China.
  • Guangxi Key Laboratory of Precision Navigation Technology and Application, Guilin University of Electronic Technology, Guilin 541004, China.
  • School of Aerospace, Harbin Institute of Technology, Shenzhen 518055, China.

ORCID-Profile: Xiuping Li, Xiyan Sun, Yuanfa Ji, Jingjing Li, Wentao Fu, Songke Zhao, Wenbin Liang, Xizi Jia, Jian Liu

Förderung

  • Guangxi Science and Technology Department
  • National Natural Science Foundation of China
  • Nanning Scientific Research and Technology Development Program
  • Guangxi Natural Science Foundation
  • Guangxi Science and Technology Program

Zitieren

APA

Li, X., Sun, X., Ji, Y., Li, J., Fu, W., Zhao, S., Liang, W., Jia, X., & Liu, J. (2026). Physics-Guided Multi-GSO Spectral Filtering for Degradation-Aware Automotive Radar Point-Cloud Detection. Sensors, 26(15), 4714. https://doi.org/10.3390/s26154714

Vancouver

Li X, Sun X, Ji Y, Li J, Fu W, Zhao S, et al. Physics-Guided Multi-GSO Spectral Filtering for Degradation-Aware Automotive Radar Point-Cloud Detection. Sensors. 2026;26(15):4714. doi:10.3390/s26154714

BibTeX
@article{Li2026,
  author = {Li, Xiuping and Sun, Xiyan and Ji, Yuanfa and Li, Jingjing and Fu, Wentao and Zhao, Songke and Liang, Wenbin and Jia, Xizi and Liu, Jian},
  title = {Physics-Guided Multi-GSO Spectral Filtering for Degradation-Aware Automotive Radar Point-Cloud Detection},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {15},
  pages = {4714},
  doi = {10.3390/s26154714}
}
Kostenlos starten →