Paper Detail
Haidong Huang, Xixin Zhao, Yaohua Zhou, Jiayu Song, Jiayi Zhang, Jun Ma, Haiyue Zhu, Xiaocong Li
Diffusion models excel at capturing multi-modal action distributions in robot imitation learning. However, in multi-task and long-horizon scenarios, monolithic architectures lack structural generalization capabilities, suffering from gradient conflicts between distinct semantic sub-stages. While pure data-driven Mixture-of-Experts (MoE) methods introduce labor division, they frequently trigger routing collapse, and instantiating full-scale experts causes parameter explosion and high expansion costs. To address these issues, we propose Concept-prior Routed Diffusion Experts (CoRDE), a structure-guided variational distillation framework. CoRDE extracts semantic distributions from a frozen concept encoder to guide the variational posterior responsibility via a learnable soft mapping matrix. This mechanism introduces an entropy-controlled responsibility inference process that encourages confident routing under reliable semantic predictions while preserving the stochastic diffusion term for behavioral diversity. To overcome parameter inflation, CoRDE employs a parameter-efficient expert pool using Low-Rank Adaptation (LoRA) on a shared frozen backbone. Theoretical analysis shows that the mixture score discrepancy is bounded by responsibility-weighted local expert errors, supporting high-fidelity generation under low-rank expert adaptation. Empirical evaluations confirm that, compared to existing baselines, CoRDE systematically reduces routing collapse, forming robust, semantically aligned expert allocations while achieving superior action quality and incremental learning efficiency.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@article{huang2026corde,
title = {CoRDE: Concept-Prior Routed Diffusion Experts for Structural Generalization in Robot Manipulation},
author = {Haidong Huang and Xixin Zhao and Yaohua Zhou and Jiayu Song and Jiayi Zhang and Jun Ma and Haiyue Zhu and Xiaocong Li},
year = {2026},
abstract = {Diffusion models excel at capturing multi-modal action distributions in robot imitation learning. However, in multi-task and long-horizon scenarios, monolithic architectures lack structural generalization capabilities, suffering from gradient conflicts between distinct semantic sub-stages. While pure data-driven Mixture-of-Experts (MoE) methods introduce labor division, they frequently trigger routing collapse, and instantiating full-scale experts causes parameter explosion and high expansion co},
url = {https://arxiv.org/abs/2606.21935},
keywords = {cs.RO},
eprint = {2606.21935},
archiveprefix = {arXiv},
}
{}