Flow2Act: Integrating Agglomerative Perception with One-step Action Generation for Robotic Manipulation

Sen Wang1, Le Wang1*, Hongcheng Huo2, Sanping Zhou1, Kun Xia1, Gang Hua3,
1Xi’an Jiaotong University 2University of Toronto 3Amazon Alexa AI
*Indicates Corresponding Author
Teaser result

Flow2Act is a robotic manipulation framework that integrates agglomerative perception with one-step action generation. It leverages a multi-teacher visual backbone, a MeanFlow policy based on an interval-averaged velocity field, and a curriculum-based region-aware mechanism to achieve high manipulation accuracy, strong generalization, and efficient real-time inference simultaneously.

Abstract

Developing generalizable robotic policies that balance inference efficiency, manipulation accuracy, and robustness remains a formidable challenge. Existing vision-language-action models demand prohibitive data scales, while keyframe-based approaches struggle to reconcile the expressivity of generative models with the latency of iterative sampling. To address this trilemma, we present Flow2Act, a unified framework that integrates agglomerative perception with a deterministic one-step generative policy. Unlike prior methods relying on separate semantic encoders or iterative diffusion processes, our approach introduces three key innovations. First, we employ an agglomerative multi-teacher visual backbone that distills complementary strengths from diverse foundation models, capturing semantics, spatial structure, and segmentation to yield robust representations without task-specific pretraining. Second, we propose a conditional MeanFlow policy that parameterizes the interval-averaged velocity field. This formulation enables genuine single-step action generation, eliminating the discretization errors and computational overhead inherent in ODE-based flow matching. Third, we devise a curriculum region-aware mechanism via a Spatial-Grounded State Space Model, which progressively shifts attention from global flow stability to fine-grained contact precision. We evaluate Flow2Act on challenging simulation benchmarks and real-world robotic tasks, demonstrating significant gains in policy performance, robustness to environmental perturbations, and cross-task real-world applicability.

Overview of Flow2Act

Results across Benchmarks

Results on RLBench

Real World Deployment

Failure Cases

We analyze all 95 failed trials in the real-world Generalization Setting. Three primary failure causes are identified: severe occlusion during local refinement (43/95, 45.3%), depth sensor artifact or missing point cloud (28/95, 29.5%), and kinematic singularity or planner-level reachability issue (24/95, 25.3%).

BibTeX

@article{wangflow2act2025,
      title={Integrating Agglomerative Perception with One-step Action Generation for Robotic Manipulation},
      author={Sen Wang, Le Wang, Hongcheng Huo, Sanping Zhou, Kun Xia, Gang Hua},
      year={2025},
}