Hierarchical Policy Learning via Spectral Decomposition
Abstract
In this paper, we identify a semantic decomposition in robot action sequences, separating task-level motion intent from execution-level refinements. By analyzing actions in the spectral domain using the discrete cosine transform (DCT), we observe that low-frequency components capture global motion trajectories, while high-frequency components encode precise timing, alignment, and contact behaviors. Motivated by this structure, we propose Causal Spectral Policy (CSP), which models action generation as a causal coarse-to-fine process: coarse motion is predicted from observation and language, and fine corrections are generated conditionally on the realized trajectory. Across simulation and real-world evaluations, CSP consistently outperforms strong baselines on precision-sensitive manipulation tasks. Additionally, we propose human-inspired teleoperation noise injection as a data augmentation method, under which our approach demonstrates strong robustness to noisy demonstrations.
Cite
@article{arxiv.2606.29570,
title = {Hierarchical Policy Learning via Spectral Decomposition},
author = {Shuxin Cao and Liquan Wang and Walker Byrnes and Yiye Chen and Yilun Du and Animesh Garg},
journal= {arXiv preprint arXiv:2606.29570},
year = {2026}
}