Research map / Reinforcement learning / Visual and representation RL
Latent Adaptation of Foundation Policies for Sim-to-Real Transfer
Longchao Da, T Pranav Kutralingam, Lirong Xiang, Hua Wei
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 |
|---|---|---|---|
| add the latent adapter g_theta on top of initial alignment and MetaDynamic signature (ablation) | Found-adapt · Walker friction F1 and gravity tasks | episode return | Improved |
| combine initial alignment and dynamics signature without adapter (F(init,dyna)) vs either alone (ablation) | Found-adapt · Walker friction F1 tasks | episode return | Hurt |
| adapt latent z via refinement steps (z_src to z_refined) | Found-adapt · Walker Stand task | return in E_real: 383.51 -> 553.86 | Improved |
| corrupt target-domain data by dropping transitions (under 50 percent) (ablation) | Found-adapt · Walker Stand G1 | return | No clear effect |
| corrupt target data by zero-masking transitions (ablation) | Found-adapt · Walker Stand G1 | return | Hurt |
| add Found-adapt over Direct-Transfer on Stand under G1 | Foundation policy · Walker Stand (G1 gravity) | average return: 494.24 -> 562.72 | Improved |
| add Found-adapt over Direct-Transfer on Walk under G4 | Foundation policy · Walker Walk (G4 gravity) | average return: 33.03 -> 42.57 | Improved |
| add Found-adapt over Direct-Transfer on Stand under G2 gravity | Foundation policy · Walker Stand (G2 gravity) | average return: 222.49 -> 231.75 | No clear effect |
Closest papers by meaning
- In-Context Policy Adaptation via Cross-Domain Skill Diffusion (AAAI 2025 · similarity 0.88)
- Efficient Test-Time Adaptation For Robot Policies (NeurIPS 2026 · similarity 0.88)
- ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Foundation Policies with Hilbert Representations (ICML 2024 · similarity 0.88)
- World Models as Reference Trajectories for Rapid Motor Adaptation (NeurIPS 2025 · similarity 0.88)
- Reasoning with Latent Diffusion in Offline Reinforcement Learning (ICLR 2024 · similarity 0.88)
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.
