Abstract
Generalization in multi-arm collaboration can be studied as composing familiar atomic skills in new ways across arms. However, existing evaluations offer limited insight into which training and architectural choices support this ability under different coordination requirements. We introduce \textbf{ACG-Bench}, a benchmark for \emph{Arm-wise Compositional Generalization} that provides a common testbed for studying skill recomposition in dual-arm policies. It contains 23 task--condition pairs across 8 task families, with 6 in-domain conditions and 17 unseen compositions covering reordering, synchronization, their combination, and cross-task composition. All methods receive the same per-arm atomic prompts, and success requires achieving the task goal while satisfying physical milestones and specified order or timing constraints. Using $π_{0.5}$ as a common vision-language-action backbone, we compare representative data-augmentation and architectural strategies with shared source data and a common evaluation protocol. Our architectural study examines arm-token grouping, skill-specific LoRA adapters (SkillLoRA), and arm-wise attention (AWA), highlighting the complementarity of skill-conditioned parameters and attention structure. Combining these choices yields \textbf{AE-VLA}, which achieves 21.53\% generalization success in simulation, compared with 2.94\% for Single $π_{0.5}$, 3.06\% for MA-VLA, and 5.53\% for two independently controlled $π_{0.5}$ policies. On physical SO101 robots, AE-VLA reaches 39.00\% mean success across five unseen conditions, compared with 10.00\% for the strongest baseline. These findings provide empirical guidance for designing dual-arm policies that generalize beyond fixed training routines.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Zhang, Z., Ran, B., Wu, Y., Zhang, Z., Wang, Y., Jiang, J., Xiao, J., Shi, W., Kang, L., Qin, Y., Yin, Z., Wang, L., & Lu, H. (2026). Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models. https://omanscience.com/en/articles/arm-wise-compositional-generalization-in-dual-arm-vision-language-action-models
MLA 9
Zhang, Zaibin, et al. "Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models." https://omanscience.com/en/articles/arm-wise-compositional-generalization-in-dual-arm-vision-language-action-models.
Chicago (author–date)
Zhang, Zaibin, Binghao Ran, Yuhan Wu, Zhongbo Zhang, Yifan Wang, Junwei Jiang, Junlan Xiao, Wangcheng Shi, Li Kang, Yiran Qin, Zhenfei Yin, Lijun Wang, and Huchuan Lu. 2026. "Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models." https://omanscience.com/en/articles/arm-wise-compositional-generalization-in-dual-arm-vision-language-action-models.
Harvard
Zhang, Z., Ran, B., Wu, Y., Zhang, Z., Wang, Y., Jiang, J., Xiao, J., Shi, W., Kang, L., Qin, Y., Yin, Z., Wang, L. and Lu, H. (2026) 'Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models', Available at: https://omanscience.com/en/articles/arm-wise-compositional-generalization-in-dual-arm-vision-language-action-models.
Vancouver
Zhang Z, Ran B, Wu Y, Zhang Z, Wang Y, Jiang J, et al. Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models. https://omanscience.com/en/articles/arm-wise-compositional-generalization-in-dual-arm-vision-language-action-models
IEEE
Z. Zhang, B. Ran, Y. Wu, Z. Zhang, Y. Wang, J. Jiang, J. Xiao, W. Shi, L. Kang, Y. Qin, Z. Yin, L. Wang, and H. Lu, "Arm-wise Compositional Generalization in Dual-Arm Vision-Language-Action Models," https://omanscience.com/en/articles/arm-wise-compositional-generalization-in-dual-arm-vision-language-action-models.