EUVIMEDEUVIMED entdecken
Crossref

Structured Deep Bilinear Koopman Modeling for Contraction-Based Control of Free-Flying Space Robots

Renhao Mao, Tao Meng, Zhonglin Zuo, Kun Wang, Dexin Zhang

IEEE Transactions on Aerospace and Electronic Systems · 2026 · 62 · 7562-7579

DOI: 10.1109/taes.2026.3671766 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Aerospace and Electronic Systems, 2026. Verfasst von Renhao Mao und 4 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben. Gefördert von: Zhejiang Provincial Natural Science Foundation of China, China Postdoctoral Science Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This article addresses the problem of robust end-effector pose control for micro–nano free-flying space robots operating in proximity to targets. To tackle this issue, we propose a hierarchical control framework that integrates trajectory generation, data-driven modeling, and contraction-based controller design. First, reference trajectories are generated via coordinate transformations between the current and target poses, leveraging reaction null-space techniques to prioritize end-effector pose accuracy while optimally preserving target observation through base-mounted camera orientation maintenance. Second, we introduce a structured deep bilinear Koopman framework that embeds domain-specific kinematic priors, enabling linear representations of the underlying nonlinear dynamics. This architecture achieves superior modeling accuracy and generalization over conventional Koopman approaches and facilitates efficient controller synthesis via its linear embeddings. Building on this formulation, we design a contraction-based feedback controller within the lifted Koopman space, ensuring stable pose tracking while mitigating the dependence on prior knowledge of the nonlinear dynamics for contraction metric construction. This method avoids computationally prohibitive metric searches in nonlinear regimes while providing formal stability guarantees. Simulations conducted in MuJoCo demonstrate the effectiveness of the proposed framework.

Förderung

  • Zhejiang Provincial Natural Science Foundation of China
  • China Postdoctoral Science Foundation

Zitieren

APA

Mao, R., Meng, T., Zuo, Z., Wang, K., & Zhang, D. (2026). Structured Deep Bilinear Koopman Modeling for Contraction-Based Control of Free-Flying Space Robots. IEEE Transactions on Aerospace and Electronic Systems, 62, 7562-7579. https://doi.org/10.1109/taes.2026.3671766

Vancouver

Mao R, Meng T, Zuo Z, Wang K, Zhang D. Structured Deep Bilinear Koopman Modeling for Contraction-Based Control of Free-Flying Space Robots. IEEE Transactions on Aerospace and Electronic Systems. 2026;62:7562-7579. doi:10.1109/taes.2026.3671766

BibTeX
@article{Mao2026,
  author = {Mao, Renhao and Meng, Tao and Zuo, Zhonglin and Wang, Kun and Zhang, Dexin},
  title = {Structured Deep Bilinear Koopman Modeling for Contraction-Based Control of Free-Flying Space Robots},
  journal = {IEEE Transactions on Aerospace and Electronic Systems},
  year = {2026},
  volume = {62},
  pages = {7562--7579},
  doi = {10.1109/taes.2026.3671766}
}
Kostenlos starten →