Research map / Vision-language and multimodal / Robot action and navigation
MAPS: Preserving Vision-Language Representations via Module-Wise Proximity Scheduling for Better Vision-Language-Action Generalization
Chengyue Huang, Mellon M. Zhang, Robert Azarcon, Glen Chou, Zsolt Kira
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 MAPS module-wise proximity scheduling | MiniVLA-OFT · SimplerEnv Avg. OOD | success rate (%): 8.9 -> 35.8 | Improved |
| add MAPS | MiniVLA-OFT · SimplerEnv Avg. ID | success rate (%): 13.5 -> 30.0 | Improved |
| add MAPS | MiniVLA-VQ · SimplerEnv | success rate (%): ID 44.0 -> 42.5; OOD 32.9 -> 34.4 | Mixed |
| add MAPS | OpenVLA-OFT · SimplerEnv Avg. OOD | success rate (%): 8.7 -> 17.1 | Improved |
| add MAPS | MiniVLA-OFT · CALVIN ABC->D | avg length: +0.7 | Improved |
| add MAPS | OpenVLA-OFT · CALVIN ABC->D | avg length: +0.1 | Improved |
| add MAPS | MiniVLA-VQ · LIBERO OOD avg | success rate (%): 0.0 -> 4.75 | Improved |
| add MAPS | OpenVLA-OFT · LIBERO OOD avg | success rate (%): 5.0 -> 5.5 | No clear effect |
| add MAPS | VLA-Adapter · LIBERO OOD avg | success rate (%): 2.5 -> 4.75 | Improved |
| add MAPS | MiniVLA-OFT · Franka real-world avg OOD | success rate (%): 22.5 -> 52.5 | Improved |
| freeze the language backbone (ablation) | OpenVLA-style VLA (MiniVLA-OFT) · SimplerEnv | success rate (%): near-zero | Hurt |
| freeze the vision encoder (ablation) | MiniVLA-OFT · SimplerEnv | success rate (%): ID +7-17%, OOD +7-25% | Improved |
Closest papers by meaning
- EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models (NeurIPS 2025 · similarity 0.92)
- CogVLA: Cognition-Aligned Vision-Language-Action Models via Instruction-Driven Routing & Sparsification (NeurIPS 2025 · similarity 0.91)
- Large Vision-Language Models Get Lost in Attention (ICML 2026 · similarity 0.90)
- From Pixels to Words -- Towards Native Vision-Language Primitives at Scale (ICLR 2026 · similarity 0.90)
- Not All Features Are Created Equal: A Mechanistic Study of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- A Stitch in Time Saves Nine: Small VLM is a Precise Guidance for Accelerating Large VLMs (CVPR 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.
