Compose Your Policies! Improving Diffusion-based or Flow-based Robot Policies via Test-time Distribution-level Composition

Jiahang Cao, Yize Huang, Hanzhong Guo, Qiang Zhang, Rui Zhang, Weijian Mai, Mu Nan, Jiaxu Wang, Hao Cheng, Jingkai SUN, Gang Han, Wen Zhao, Yijie Guo, Qihao Zheng, Xiao Li, Chunfeng Song, Ping Luo, Andrew Luo

Keywords: Diffusion Policies, Policy Composition, Training-free

TL;DR: We present General Policy Composition, a training-free method that enhances performance by combining the distributional scores of multiple pre-trained policies via a convex combination and test-time search.

Abstract: Diffusion-based models for robotic control, including vision-language-action (VLA) and vision-action (VA) policies, have demonstrated significant capabilities. Yet their advancement is constrained by the high cost of acquiring large-scale interaction datasets. This work introduces an alternative paradigm for enhancing policy performance ***without additional model training***. Perhaps surprisingly, we demonstrate that the composed policies can exceed the performance of either parent policy. Our contribution is threefold. First, we establish a theoretical foundation showing that the convex composition of distributional scores from multiple diffusion models can yield a superior one-step functional objective compared to any individual score. A Grönwall-type bound is then used to show that this single-step improvement propagates through entire generation trajectories, leading to systemic performance gains. Second, motivated by these results, we propose General Policy Composition (GPC), a training-free method that enhances performance by combining the distributional scores of multiple pre-trained policies via a convex combination and test-time search. GPC is versatile, allowing for the plug-and-play composition of heterogeneous policies, including VA and VLA models, as well as those based on diffusion or flow-matching, irrespective of their input visual modalities. Third, we provide extensive empirical validation. Experiments on Robomimic, PushT, and RoboTwin benchmarks, alongside real-world robotic evaluations, confirm that GPC consistently improves performance and adaptability across a diverse set of tasks. Further analysis of alternative composition operators and weighting strategies offers insights into the mechanisms underlying the success of GPC. These results establish GPC as a simple yet effective method for improving control performance by leveraging existing policies.

Primary Area: applications to robotics, autonomy, planning

Submission Number: 5578