AI Paper Atlas

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

ECCV 2026 · Vision-language and multimodal › Robot action and navigation · approach: Memory and subgoal-based manipulation · 5 citations (OpenAlex)

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 changedWhereMeasuredEffect
add a second SFT epoch (ablation)Ours-8B SFT · REAL-Bench FDPsuccess rate (%): 45.8 -> 65.3Improved
add a second SFT epoch (ablation)Ours-8B SFT · REAL-Bench FODPsuccess rate (%): 30.4 -> 28.6Hurt
add a second SFT epoch (ablation)Ours-8B SFT · REAL-Bench FDOsuccess rate (%): 35.4 -> 43.8Improved
add a second SFT epoch (ablation)Ours-8B SFT · REAL-Bench SULsuccess rate (%): 36.9 -> 50.8Improved
add GSPO online RL after 1-epoch SFT (ablation)Ours-8B SFT · REAL-Bench SULsuccess rate (%): 36.9 -> 56.9Improved
add GSPO online RL after 1-epoch SFT (ablation)Ours-8B SFT · REAL-Bench FODPsuccess rate (%): 30.4 -> 33.9Improved
add GSPO online RL after 1-epoch SFT (ablation)Ours-8B SFT · REAL-Bench FDPsuccess rate (%): 45.8 -> 58.3Improved
add GSPO online RL after 1-epoch SFT (ablation)Ours-8B SFT · REAL-Bench FDOsuccess rate (%): 35.4 -> 41.7Improved
add GSPO RL instead of a second SFT epoch (ablation)Ours-8B · REAL-Bench FDPsuccess rate (%): 65.3 -> 58.3Hurt
add one epoch SFT on tool-use trajectoriesQwen3-VL-8B-Instruct · REAL-Bench FDPsuccess rate (%): 0.0 -> 45.8Improved
add one epoch SFT on tool-use trajectoriesQwen3-VL-8B-Instruct · REAL-Bench SULsuccess rate (%): 1.5 -> 36.9Improved
scale model: Qwen3-VL-235B zero-shot vs trained 8B SFT+RLQwen3-VL · REAL-Bench SULsuccess rate (%): 18.5 -> 56.9Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy