Research map / Vision-language and multimodal / Robot action and navigation
Grounding Actions in Camera Space: Observation-Centric Vision-Language-Action Policy
Tianyi Zhang, Haonan Duan, Haoran Hao, Yu Qiao, Jifeng Dai, Zhi Hou
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 |
|---|---|---|---|
| predict actions in camera frame instead of robot base (continuous) (ablation) | Dita-style VLA · ManiSkill2 (5 tasks, from scratch) | success rate: 45.2% -> 53.2% | Improved |
| predict actions in camera frame instead of robot base (discrete) (ablation) | Dita-style VLA · ManiSkill2 (5 tasks, from scratch) | success rate: 38.6% -> 52.4% | Improved |
| camera-frame targets on PickSingleEGAD (continuous) (ablation) | Dita-style VLA · ManiSkill2 PickSingleEGAD | success rate: 48.0% -> 48.0% | No clear effect |
| camera-frame targets, fixed camera real robot (ablation) | Dita-style VLA · Real robot, 8 tasks, fixed camera | success rate: 66.3% -> 77.5% | Improved |
| camera-frame targets under camera perturbations (ablation) | Dita-style VLA · Real robot, 8 tasks, camera perturbations | success rate: 61.3% -> 73.8% | Improved |
| camera-frame targets, zero-shot novel camera (ablation) | Dita-style VLA · Real robot, 15 tasks, zero-shot novel camera | success rate: 41.3% -> 54.0% | Improved |
| camera-frame targets, training camera 1 (ablation) | Dita-style VLA · Real robot, 15 tasks, training camera 1 | success rate: 58.0% -> 68.0% | Improved |
Closest papers by meaning
- ATI-VLA: Action-Centric Predictive Vision–Language–Action Models via Actionable Alignment Then Adaptive Injection (NeurIPS 2026 · similarity 0.92)
- Beyond World-Frame Action Heads: Motion-Centric Action Frames for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.92)
- CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (CVPR 2025 · similarity 0.91)
- Beyond Data Scaling: Representation-Centric Pre-training for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.91)
- VLA-REPLICA: A Low-Cost, Reproducible Benchmark for Real-World Evaluation of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching (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.
