AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

CombatVLA: An Efficient Vision-Language-Action Model for Combat Tasks in 3D Action Role-Playing Games

Peng Chen, Pi Bu, Yingyao Wang, Xinyi Wang, Ziming Wang, Jie Guo, Yingxiu Zhao, Qi Zhu, Jun Song, Siran Yang, Jiamang Wang, Bo Zheng

ICCV 2025 · Vision-language and multimodal › Robot action and navigation · approach: Embodied multimodal LLM agents · 25 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 frame-level Frames-AoT tuning (stage 2) after video-AoT (ablation)CombatVLA · CUBenchaverage accuracy: 57.27 -> 61.43Improved
add truncated AoT tuning (stage 3) after stage 2 (ablation)CombatVLA · CUBenchaverage accuracy: 61.43 -> 63.61Improved
add truncated AoT (stage 3) (ablation)CombatVLA · CUBench reasoningaccuracy: 62.86 -> 69.71Improved
add truncated AoT (stage 3) (ablation)CombatVLA · CUBenchtime per call (s): 3.73 -> 1.85Improved
add truncated AoT (stage 3) (ablation)CombatVLA · CUBench comprehensionaccuracy: 62.25 -> 60.29Hurt
add modality contrastive loss L_con (ablation)CombatVLA · CUBenchaverage accuracy: 61.58 -> 63.61Improved
add action alignment loss L_align (ablation)CombatVLA · CUBenchaverage accuracy: 61.64 -> 63.61Improved
add L_con (ablation)CombatVLA · CUBench reasoningaccuracy: 63.14 -> 69.71Improved
add L_align (ablation)CombatVLA · CUBench reasoningaccuracy: 63.71 -> 69.71Improved
add L_con (ablation)CombatVLA · CUBench gatheringaccuracy: 62.78 -> 60.83Hurt
truncate AoT output at TRUNC tokenCombatVLA · Average AoT token lengthtokens: 116.57 -> 43.10Improved
fine-tune Qwen2.5-VL-3B with AoT (CombatVLA)Qwen2.5-VL-3B · CUBenchaverage accuracy: 55.87 -> 63.61Improved

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