Research map / Vision-language and multimodal / Robot action and navigation
Exploratory, Communicative, and Deployable: Vision-Driven Embodied Agents for Open-World Mobile Manipulation
Boyu Mi, Mengchen Ma, Yifei Yao, Xing Gao, Hanqing Wang, Junting Chen, Yangzi Li, Zihou Zhu, Guohao Li, Zhenfei Yin, Tai Wang, Yao Mu and 1 more
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 a second SFT epoch (ablation) | Ours-8B SFT · REAL-Bench FDP | success rate (%): 45.8 -> 65.3 | Improved |
| add a second SFT epoch (ablation) | Ours-8B SFT · REAL-Bench FODP | success rate (%): 30.4 -> 28.6 | Hurt |
| add a second SFT epoch (ablation) | Ours-8B SFT · REAL-Bench FDO | success rate (%): 35.4 -> 43.8 | Improved |
| add a second SFT epoch (ablation) | Ours-8B SFT · REAL-Bench SUL | success rate (%): 36.9 -> 50.8 | Improved |
| add GSPO online RL after 1-epoch SFT (ablation) | Ours-8B SFT · REAL-Bench SUL | success rate (%): 36.9 -> 56.9 | Improved |
| add GSPO online RL after 1-epoch SFT (ablation) | Ours-8B SFT · REAL-Bench FODP | success rate (%): 30.4 -> 33.9 | Improved |
| add GSPO online RL after 1-epoch SFT (ablation) | Ours-8B SFT · REAL-Bench FDP | success rate (%): 45.8 -> 58.3 | Improved |
| add GSPO online RL after 1-epoch SFT (ablation) | Ours-8B SFT · REAL-Bench FDO | success rate (%): 35.4 -> 41.7 | Improved |
| add GSPO RL instead of a second SFT epoch (ablation) | Ours-8B · REAL-Bench FDP | success rate (%): 65.3 -> 58.3 | Hurt |
| add one epoch SFT on tool-use trajectories | Qwen3-VL-8B-Instruct · REAL-Bench FDP | success rate (%): 0.0 -> 45.8 | Improved |
| add one epoch SFT on tool-use trajectories | Qwen3-VL-8B-Instruct · REAL-Bench SUL | success rate (%): 1.5 -> 36.9 | Improved |
| scale model: Qwen3-VL-235B zero-shot vs trained 8B SFT+RL | Qwen3-VL · REAL-Bench SUL | success rate (%): 18.5 -> 56.9 | Improved |
Closest papers by meaning
- OWMM-Agent: Open World Mobile Manipulation With Multi-modal Agentic Data Synthesis (NeurIPS 2025 · similarity 0.90)
- CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot Manipulation (ICML 2026 · similarity 0.89)
- SPIN: Simultaneous Perception Interaction and Navigation (CVPR 2024 · similarity 0.89)
- ManiTaskGen: A Comprehensive Task Generator for Benchmarking and Improving Vision-Language Agents on Embodied Decision-Making (NeurIPS 2026 · similarity 0.89)
- Position: Scaling Simulation is Neither Necessary Nor Sufficient for In-the-Wild Robot Manipulation (ICML 2024 · similarity 0.88)
- MoManipVLA: Transferring Vision-language-action Models for General Mobile Manipulation (CVPR 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.
