EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Bending–Rebound Angle Inspection System and Method for Cigarette-Box Packaging Materials

Pan He, Ming Lin, Jiangshan Hu, Jiguang Hu, Xiaoli Zhu, Wei Chu

IEEE Access · 2026 · 14 · 125940-125954

DOI: 10.1109/access.2026.3718319 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Pan He und 5 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Huanghelou Science Park Fund.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The bending-rebound angle of cigarette-box packaging materials is an important mechanical indicator for assessing material resilience, adhesive integrity, and overall packaging quality. To achieve accurate and repea measurement, this paper presents a robotic inspection system for automated bending-rebound angle evaluation. The proposed platform integrates a pressing module, an industrial robot, a linear guide rail, and a 3D scanner to perform coordinated loading and post-deformation point-cloud acquisition. Two key technical challenges are addressed. First, to improve the consistency of robotic pressing, a hierarchical robot error compensation framework is developed by combining a model-based stiffness error correction layer with a physics-informed Gaussian process model for non-geometric residual compensation. This design enables more accurate end-effector positioning under contact conditions. Second, to robustly estimate the rebound angle from reflective and noisy point clouds, an uncertainty-aware robust two-plane segmentation (URTPS) method is proposed. By incorporating uncertainty-normalized residuals, soft probabilistic assignment, and posterior robustness reweighting, the method achieves stable plane fitting and reliable normal estimation in the presence of perturbations and outliers. Experimental results demonstrate that the proposed compensation framework reduces the average actual positioning error to 0.0620 mm. In synthetic noisy point-cloud tests, URTPS keeps the angle estimation error below 0.15° under severe noise and outlier contamination. In system-level accuracy evaluation, using CMM-measured 3D-printed reference parts, the proposed pipeline achieves an average MAE of 0.12° and a maximum per-angle MAE of 0.14°. Real-sample measurements further show high repeatability, with standard deviations ranging from 0.01° to 0.13°.

Förderung

  • Huanghelou Science Park Fund

Zitieren

APA

He, P., Lin, M., Hu, J., Hu, J., Zhu, X., & Chu, W. (2026). Bending–Rebound Angle Inspection System and Method for Cigarette-Box Packaging Materials. IEEE Access, 14, 125940-125954. https://doi.org/10.1109/access.2026.3718319

Vancouver

He P, Lin M, Hu J, Hu J, Zhu X, Chu W. Bending–Rebound Angle Inspection System and Method for Cigarette-Box Packaging Materials. IEEE Access. 2026;14:125940-125954. doi:10.1109/access.2026.3718319

BibTeX
@article{He2026,
  author = {He, Pan and Lin, Ming and Hu, Jiangshan and Hu, Jiguang and Zhu, Xiaoli and Chu, Wei},
  title = {Bending–Rebound Angle Inspection System and Method for Cigarette-Box Packaging Materials},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {125940--125954},
  doi = {10.1109/access.2026.3718319}
}
Kostenlos starten →