FocalPolicy: Frequency-Optimized Chunking and Locally Anchored Flow Matching for Coherent Visuomotor Policy

TL;DR

FocalPolicy is a novel visuomotor policy that enhances long-horizon task coherence by combining frequency-optimized chunking with locally anchored flow matching, balancing precision and foresight.

Contribution

It introduces a foresight-aware approach with a composite objective and locally anchored sampling, improving inter-chunk coherence and generalizability in visuomotor learning.

Findings

FocalPolicy outperforms existing methods in experiments.

Modules are generalizable to other baselines.

Achieves better long-horizon coherence in manipulation tasks.

Abstract

Visuomotor policies aim to learn complex manipulation tasks from expert demonstrations. However, generating smooth and coherent trajectories remains challenging, as it requires balancing proximal precision with distal foresight. Existing approaches typically focus on optimizing intra-chunk action distributions, often neglecting the inter-chunk coherence. Consequently, inter-chunk discontinuities significantly impede the learning of coherent long-horizon actions. To overcome this limitation and achieve a synergetic balance between precision and foresight, we propose FocalPolicy, a foresight-aware visuomotor policy that combines Frequency-Optimized Chunking with Locally Anchored flow matching. We introduce a foresight composite objective that supervises time-domain alignment within the proximal actions while regularizing frequency-domain structure over multiple future action chunks to improve cross-chunk coherence. To efficiently learn complex action distributions, we design locally anchored sampling to enhance target signal propagation efficiency during consistency flow matching training. Extensive experiments demonstrate that FocalPolicy outperforms existing approaches and confirm the generalizability of our modules to other baselines. Project website: https://focalpolicy.github.io/