Research map / Vision-language and multimodal / Robot action and navigation
Sim2Real VLA: Zero-Shot Generalization of Synthesized Skills to Realistic Manipulation
Runyi Zhao, Sheng Xu, Ruixing Jin, Yueci Deng, Yunxin Tai, Kui Jia, Guiliang Liu
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 |
|---|---|---|---|
| use arm-decoupled control modules instead of joint learning (ablation) | Sim2Real-VLA · Items Hand-Over and Place (real) | real-world success rate: 0.15 -> 0.40 | Improved |
| use arm-decoupled control modules instead of joint learning (ablation) | Sim2Real-VLA · Items Hand-Over and Place (sim) | simulation success rate: 0.32 -> 0.62 | Improved |
| use arm-decoupled control modules instead of joint learning (ablation) | Sim2Real-VLA · Single-Arm Water Pouring (real) | real-world success rate: 0.75 -> 0.85 | Improved |
| lengthen affordance chain from K=1 to K=3 (ablation) | Sim2Real-VLA · Single-Arm Water Pour | real-world successes out of 20: 17/20 -> 11/20 | Hurt |
| lengthen affordance chain from K=1 to K=3 (ablation) | Sim2Real-VLA · Dual-Arm Water Pour | real-world successes out of 20: 16/20 -> 8/20 | Hurt |
| lengthen affordance chain from K=1 to K=3 (ablation) | Sim2Real-VLA · Basket Pick-and-Place | real-world successes out of 20: 9/20 -> 4/20 | Hurt |
| lengthen affordance chain from K=1 to K=3 (ablation) | Sim2Real-VLA · Items Hand-Over and Place | real-world successes out of 20: 8/20 -> 4/20 | Hurt |
| train on synthetic data with Sim2Real-VLA instead of best baseline pi0-FAST | VLA policy fine-tuned on simulator data · Single-Arm Water Pour (real) | real-world successes out of 20: 11/20 -> 17/20 | Improved |
| train on synthetic data with Sim2Real-VLA instead of GR00T N1.5 | VLA policy fine-tuned on simulator data · Dual-Arm Water Pour (real) | real-world successes out of 20: 7/20 -> 16/20 | Improved |
| change background texture in real environment | Sim2Real-VLA · Single-Arm Water Pour | real-world successes out of 20: 17/20 -> 17/20 | No clear effect |
| change object features in real environment | Sim2Real-VLA · Table Rearrangement | real-world successes out of 20: 16/20 -> 14/20 | Hurt |
| change table texture in real environment | Sim2Real-VLA · Items Hand-Over and Place | real-world successes out of 20: 8/20 -> 6/20 | Hurt |
Closest papers by meaning
- VLA-REPLICA: A Low-Cost, Reproducible Benchmark for Real-World Evaluation of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.91)
- VLA-ATTC: Adaptive Test-Time Compute for VLA Models with Relative Action Critic Model (ICML 2026 · similarity 0.91)
- NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- Exploring the Limits of Vision-Language-Action Manipulation in Cross-task Generalization (NeurIPS 2025 · similarity 0.90)
- VLA-OS: Structuring and Dissecting Planning Representations and Paradigms in Vision-Language-Action Models (NeurIPS 2025 · similarity 0.90)
- VideoVLA: Video Generators Can Be Generalizable Robot Manipulators (NeurIPS 2025 · similarity 0.90)
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.
