EUVIMEDEUVIMED entdecken
Crossref

EDIL: An End-to-End Decoupled Imitation Learning Method for Stable Long-Horizon Bimanual Manipulation

Zhipeng Wang, Chaoyun Yang, Rong Jiang, Jinyu Zou, Gengdong Zhou, Yanmin Zhou, Bin He

IEEE Transactions on Industrial Informatics · 2026-07 · 22 (7) · 6605-6616

DOI: 10.1109/tii.2026.3679954 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Industrial Informatics, 2026-07. Verfasst von Zhipeng Wang und 6 weiteren Autor:innen. Die Arbeit nennt 33 Literaturangaben. Gefördert von: National Key Research and Development Program of China, National Natural Science Foundation of China, Shanghai Municipal Science and Technology Major Project, China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As a critical capability for automating complex industrial tasks like precision assembly, bimanual fine-grained manipulation has become a key area of research in robotics, where end-to-end imitation learning has emerged as a prominent paradigm. However, in long-horizon cooperative tasks, prevalent end-to-end methods suffer from an inadequate representation of critical task features, particularly those essential for fine-grained coordination between the arms and grippers. This deficiency often destabilizes the manipulation policy, leading to issues like spurious gripper activations and culminating in task failure. To address this challenge, we propose an end-to-end decoupled imitation learning (EDIL) method, which decouples the bimanual manipulation task into a multimodal arm policy for global trajectories and a temporally coordinated attentive gripper policy for fine end-effector actions. The arm policy leverages a Transformer-based encoder–decoder architecture to learn multimodal trajectories from expert demonstrations. The gripper policy leverages cross-attention to facilitate implicit, dynamic feature sharing between the arms, integrating sequential state history and visual data to ensure cooperative stability. We evaluated EDIL on three challenging long-horizon manipulation tasks. Experimental results demonstrate that our method significantly outperforms state-of-the-art approaches, particularly on more complex subtasks, showcasing its robustness and effectiveness.

Förderung

  • National Key Research and Development Program of China
  • National Natural Science Foundation of China
  • Shanghai Municipal Science and Technology Major Project, China
  • China Postdoctoral Science Foundation

Zitieren

APA

Wang, Z., Yang, C., Jiang, R., Zou, J., Zhou, G., Zhou, Y., & He, B. (2026). EDIL: An End-to-End Decoupled Imitation Learning Method for Stable Long-Horizon Bimanual Manipulation. IEEE Transactions on Industrial Informatics, 22(7), 6605-6616. https://doi.org/10.1109/tii.2026.3679954

Vancouver

Wang Z, Yang C, Jiang R, Zou J, Zhou G, Zhou Y, et al. EDIL: An End-to-End Decoupled Imitation Learning Method for Stable Long-Horizon Bimanual Manipulation. IEEE Transactions on Industrial Informatics. 2026;22(7):6605-6616. doi:10.1109/tii.2026.3679954

BibTeX
@article{Wang2026,
  author = {Wang, Zhipeng and Yang, Chaoyun and Jiang, Rong and Zou, Jinyu and Zhou, Gengdong and Zhou, Yanmin and He, Bin},
  title = {EDIL: An End-to-End Decoupled Imitation Learning Method for Stable Long-Horizon Bimanual Manipulation},
  journal = {IEEE Transactions on Industrial Informatics},
  year = {2026},
  volume = {22},
  number = {7},
  pages = {6605--6616},
  doi = {10.1109/tii.2026.3679954}
}
Kostenlos starten →