Research map / Vision-language and multimodal / Robot action and navigation
Towards Generalizable Robotic Manipulation in Dynamic Environments
Heng Fang, Shangru Li, Shuhan Wang, Xuanyang Xi, Dingkang Liang, Xiang Bai
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 historical optical flow (ablation) | OpenVLA-OFT (Qwen3-VL) · DOMINO@0.1 Level 1 | SR: 10.86 -> 11.71 | Improved |
| add auxiliary future prediction (N=2) (ablation) | PUMA with Hist. Flow · DOMINO@0.1 Level 1 | SR: 11.71 -> 14.80 | Improved |
| use optical flow instead of raw historical frames (ablation) | PUMA (N=2) · DOMINO@0.1 Level 1 | SR: 8.15 -> 14.80 | Improved |
| extend prediction horizon from N=2 to N=4 (ablation) | PUMA · DOMINO@0.1 Level 1 | SR: 14.80 -> 17.20 | Improved |
| extend prediction horizon from N=4 to N=6 (ablation) | PUMA · DOMINO@0.1 Level 1 | SR: 17.20 -> 13.62 | Hurt |
| use history length h=4 instead of h=1 (ablation) | PUMA · DOMINO@0.1 Level 1 | SR: 13.00 -> 17.20 | Improved |
| co-train with static data (ablation) | PUMA · DOMINO@0.1 Level 1 | SR: 14.80 -> 19.71 | Improved |
| co-train with static data (ablation) | OpenVLA-OFT (Qwen3-VL) · DOMINO@0.1 Level 1 | SR: 10.86 -> 12.89 | Improved |
| fine-tune on dynamic data instead of zero-shot from static | pi0.5 · DOMINO@0.1 | SR: 7.5 -> 9.6 | No clear effect |
| move from static to dynamic environment | pi0.5 · RoboTwin 2.0 vs DOMINO@0.1 | SR: 44.8 -> 7.5 | Hurt |
| inject ground-truth future trajectory without history | OpenVLA-OFT · DOMINO@0.1 Level 1 | SR: 9.06 -> 10.33 | No clear effect |
| inject ground-truth future trajectory without history | OpenVLA-OFT · DOMINO@0.1 Level 1 | MS: 24.06 -> 32.00 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VLA-Adapter — VLA-Adapter: An Effective Paradigm for Tiny-Scale Vision-Language-Action Model (AAAI 2026) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Exploring the Limits of Vision-Language-Action Manipulation in Cross-task Generalization (NeurIPS 2025 · similarity 0.91)
- ActionUNet: Improving Robustness of VLA Models with Efficient Multi-scale Fine-tuning (NeurIPS 2026 · similarity 0.91)
- DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (NeurIPS 2025 · similarity 0.91)
- OmniManip: Towards General Robotic Manipulation via Object-Centric Interaction Primitives as Spatial Constraints (CVPR 2025 · similarity 0.91)
- VideoVLA: Video Generators Can Be Generalizable Robot Manipulators (NeurIPS 2025 · similarity 0.91)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · 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.
