Abstract
Action chunking is widely used for action generation and execution in Vision-Language-Action (VLA) policies, yet existing approaches commonly use a fixed action horizon. During a rollout, different task stages may require different levels of action continuity, control precision, and closed-loop feedback, making a fixed horizon unable to accommodate changing control requirements. We propose \textbf{GeoAAC}, a geometry-based adaptive action chunking method for flow-based VLA policies that adjusts the action horizon according to the reliability of the current action prediction. We show that the geometry of Flow Matching denoising trajectories provides process-level information for characterizing prediction reliability, with geometric variation across action prefixes remaining positively correlated with predictive uncertainty. GeoAAC uses this prefix-wise geometry to construct a horizon-wise geometric profile and adaptively determine the action horizon from a single generation without additional training. Experiments with GR00T N1.5 and π0.5 on LIBERO, LIBERO-Pro, RoboCasa365, and real-world manipulation tasks show consistent improvements over fixed-action-horizon baselines and existing adaptive methods, including up to 8.7 percentage points in simulation and an increase in average real-world success rate from 53.3\% to 74.4\%.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Chen, X., Chen, S., Ding, Y., Liu, J., Wang, G., Ye, W., Shen, H. T., & Bin, Y. (2026). GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies. https://omanscience.com/en/articles/geoaac-geometry-based-adaptive-action-chunking-from-denoising-trajectories-in-vla-policies
MLA 9
Chen, Xin, et al. "GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies." https://omanscience.com/en/articles/geoaac-geometry-based-adaptive-action-chunking-from-denoising-trajectories-in-vla-policies.
Chicago (author–date)
Chen, Xin, Sen Chen, Yujuan Ding, Jian Liu, Guoqing Wang, Wei Ye, Heng Tao Shen, and Yi Bin. 2026. "GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies." https://omanscience.com/en/articles/geoaac-geometry-based-adaptive-action-chunking-from-denoising-trajectories-in-vla-policies.
Harvard
Chen, X., Chen, S., Ding, Y., Liu, J., Wang, G., Ye, W., Shen, H. T. and Bin, Y. (2026) 'GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies', Available at: https://omanscience.com/en/articles/geoaac-geometry-based-adaptive-action-chunking-from-denoising-trajectories-in-vla-policies.
Vancouver
Chen X, Chen S, Ding Y, Liu J, Wang G, Ye W, et al. GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies. https://omanscience.com/en/articles/geoaac-geometry-based-adaptive-action-chunking-from-denoising-trajectories-in-vla-policies
IEEE
X. Chen, S. Chen, Y. Ding, J. Liu, G. Wang, W. Ye, H. T. Shen, and Y. Bin, "GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies," https://omanscience.com/en/articles/geoaac-geometry-based-adaptive-action-chunking-from-denoising-trajectories-in-vla-policies.