MoE-DP: An MoE-Enhanced Diffusion Policy for Robust Long-Horizon Robotic Manipulation with Skill Decomposition and Failure Recovery
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911253486632960 |
|---|---|
| author | Cheng, Baiye Liang, Tianhai Huang, Suning Shao, Maanping Zhang, Feihong Xu, Botian Xue, Zhengrong Xu, Huazhe |
| author_facet | Cheng, Baiye Liang, Tianhai Huang, Suning Shao, Maanping Zhang, Feihong Xu, Botian Xue, Zhengrong Xu, Huazhe |
| contents | Diffusion policies have emerged as a powerful framework for robotic visuomotor control, yet they often lack the robustness to recover from subtask failures in long-horizon, multi-stage tasks and their learned representations of observations are often difficult to interpret. In this work, we propose the Mixture of Experts-Enhanced Diffusion Policy (MoE-DP), where the core idea is to insert a Mixture of Experts (MoE) layer between the visual encoder and the diffusion model. This layer decomposes the policy's knowledge into a set of specialized experts, which are dynamically activated to handle different phases of a task. We demonstrate through extensive experiments that MoE-DP exhibits a strong capability to recover from disturbances, significantly outperforming standard baselines in robustness. On a suite of 6 long-horizon simulation tasks, this leads to a 36% average relative improvement in success rate under disturbed conditions. This enhanced robustness is further validated in the real world, where MoE-DP also shows significant performance gains. We further show that MoE-DP learns an interpretable skill decomposition, where distinct experts correspond to semantic task primitives (e.g., approaching, grasping). This learned structure can be leveraged for inference-time control, allowing for the rearrangement of subtasks without any re-training.Our video and code are available at the https://moe-dp-website.github.io/MoE-DP-Website/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_05007 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MoE-DP: An MoE-Enhanced Diffusion Policy for Robust Long-Horizon Robotic Manipulation with Skill Decomposition and Failure Recovery Cheng, Baiye Liang, Tianhai Huang, Suning Shao, Maanping Zhang, Feihong Xu, Botian Xue, Zhengrong Xu, Huazhe Robotics Diffusion policies have emerged as a powerful framework for robotic visuomotor control, yet they often lack the robustness to recover from subtask failures in long-horizon, multi-stage tasks and their learned representations of observations are often difficult to interpret. In this work, we propose the Mixture of Experts-Enhanced Diffusion Policy (MoE-DP), where the core idea is to insert a Mixture of Experts (MoE) layer between the visual encoder and the diffusion model. This layer decomposes the policy's knowledge into a set of specialized experts, which are dynamically activated to handle different phases of a task. We demonstrate through extensive experiments that MoE-DP exhibits a strong capability to recover from disturbances, significantly outperforming standard baselines in robustness. On a suite of 6 long-horizon simulation tasks, this leads to a 36% average relative improvement in success rate under disturbed conditions. This enhanced robustness is further validated in the real world, where MoE-DP also shows significant performance gains. We further show that MoE-DP learns an interpretable skill decomposition, where distinct experts correspond to semantic task primitives (e.g., approaching, grasping). This learned structure can be leveraged for inference-time control, allowing for the rearrangement of subtasks without any re-training.Our video and code are available at the https://moe-dp-website.github.io/MoE-DP-Website/. |
| title | MoE-DP: An MoE-Enhanced Diffusion Policy for Robust Long-Horizon Robotic Manipulation with Skill Decomposition and Failure Recovery |
| topic | Robotics |
| url | https://arxiv.org/abs/2511.05007 |