Research map / Vision-language and multimodal / Robot action and navigation
Dismantling the Illusion of Vision-Language-Action Models Competence via Explicit Distributional Shifts
Xueyang Zhou, Yangming Xu, Guiyao Tie, Chaoran Hu, Bo Tao, xingwei zhao, Xiang Xiang, Pan Zhou, Lichao Sun, Yongchao Chen
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 structured Stair sampling instead of random sampling (ablation) | Pi0.5 · LIBERO-Gen Position generalization | success rate (%): 52.3 -> 64.0 | Improved |
| use structured Stair sampling instead of random sampling (ablation) | Pi0.5 · LIBERO-Gen Task generalization | success rate (%): 14.2 -> 21.2 | Improved |
| fine-tune on LIBERO-Gen structured subsets | Pi0 · LIBERO-Gen Position (k2) | success rate (%): 7.2 -> 13.2 | Improved |
| fine-tune on LIBERO-Gen structured subsets | Pi0.5 · LIBERO-Gen Position (k2) | success rate (%): 49.2 -> 64.0 | Improved |
| fine-tune on LIBERO-Gen structured subsets | Pi0 · LIBERO-Gen Task (k2) | success rate (%): 13.6 -> 13.6 | No clear effect |
| add consistency-regularized augmentation (ablation) | OpenVLA · LIBERO-Gen Language (k1) | accuracy (%): 0.0 -> 98.5 | Improved |
| add consistency-regularized augmentation (ablation) | OpenVLA · LIBERO-Gen Language (k3, DG) | accuracy (%): 39.2 -> 98.6 | Improved |
| add consistency-regularized augmentation (ablation) | OpenVLA · LIBERO-Gen Object (k2) | accuracy (%): 63.8 -> 88.0 | Improved |
| add consistency-regularized augmentation (ablation) | OpenVLA · LIBERO-Gen Background (k2) | accuracy (%): 94.0 -> 98.8 | Improved |
| add consistency-regularized augmentation (ablation) | OpenVLA · LIBERO-Gen Background (k3, DG) | accuracy (%): 99.4 -> 98.8 | No clear effect |
| fine-tune on LIBERO-Gen | OpenVLA · LIBERO-Gen Language (k1) | accuracy (%): 0.0 -> 91.4 | Improved |
| fine-tune on LIBERO-Gen | OpenVLA · LIBERO-Gen Background (k1, ID) | accuracy (%): 99.6 -> 98.8 | No clear effect |
Closest papers by meaning
- Not All Features Are Created Equal: A Mechanistic Study of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.92)
- VLA-REPLICA: A Low-Cost, Reproducible Benchmark for Real-World Evaluation of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- World–Value–Action Model: Implicit Planning for Vision–Language–Action Systems (NeurIPS 2026 · similarity 0.90)
- CofactVLA: Deconfounding Vision-Language-Action Models via Counterfactual Intervention (NeurIPS 2026 · similarity 0.89)
- EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models (NeurIPS 2025 · similarity 0.89)
- Exploring the Limits of Vision-Language-Action Manipulation in Cross-task Generalization (NeurIPS 2025 · 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.
