Research map / Label-efficient and robust learning / Domain adaptation and generalization
DADP: Domain Adaptive Diffusion Policy
Pengcheng Wang, Qinghang Liu, Haotian Lin, Yiheng Li, Guojian Zhan, Masayoshi Tomizuka, Yixiao Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase context lag delta t from 1 to infinity (ablation) | domain encoder · Walker2d | linear probe accuracy: 27.9% -> 99.3% | Improved |
| increase context lag delta t from 1 to infinity (ablation) | domain encoder · HalfCheetah | linear probe accuracy: 68.6% -> 99.9% | Improved |
| increase context lag delta t from 1 to infinity (ablation) | domain encoder · Hopper | linear probe accuracy: 9.0% -> 26.4% | Improved |
| increase context lag delta t from 1 to infinity (ablation) | domain encoder · Walker2d | reconstruction loss: 476.1 -> 3.2 | Improved |
| use better representation (delta t = inf instead of 1) in conditional policy (ablation) | conditional diffusion policy · Walker2d Seen | return: 2093 -> 3394 | Improved |
| condition policy on delta t = inf representation vs no representation (ablation) | diffusion policy · Walker2d Seen | return: 3722 -> 3394 | Hurt |
| bias prior with mixed Gaussian (Mixed DDIM, no prediction target) vs conditioning (ablation) | diffusion policy · HalfCheetah Unseen | return: 2744 -> 3012 | Improved |
| add representation shift to the diffusion prediction target (w/ Predict) (ablation) | Mixed DDIM · Walker2d Seen | return: 3356 -> 3991 | Improved |
| add representation shift to prediction target (ablation) | Mixed DDIM · HalfCheetah mastery | mastery: 84% -> 96% | Improved |
| learned unsupervised lagged representation vs supervised representation (ablation) | DADP · Walker2d Seen | return: 3991 vs 4014 | No clear effect |
| DADP vs Meta-DT on Walker2d | domain-adaptive policy · Walker2d Seen | return: 1304 -> 3999 | Improved |
| DADP vs Prompt-DT on Door (Adroit) Seen | domain-adaptive policy · Door Seen | return: 2116 -> 1428 | Hurt |
Closest papers by meaning
- Cross-Domain Policy Adaptation by Capturing Representation Mismatch (ICML 2024 · similarity 0.89)
- Act to See, See to Act: Diffusion-Driven Perception-Action Interplay for Adaptive Policies (NeurIPS 2025 · similarity 0.87)
- Prediction with Action: Visual Policy Learning via Joint Denoising Process (NeurIPS 2024 · similarity 0.87)
- In-Context Policy Adaptation via Cross-Domain Skill Diffusion (AAAI 2025 · similarity 0.87)
- One-Step Diffusion Policy: Fast Visuomotor Policies via Diffusion Distillation (ICML 2025 · similarity 0.86)
- Generalizable Domain Adaptation for Sim-and-Real Policy Co-Training (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
