AI Paper Atlas

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

From Manuals to Actions: A Unified VLA Model for Chain-of-Thought Manual Generation and Robotic Manipulation

Chenyang Gu, Jiaming Liu, Hao Chen, Runzhong Huang, Qingpo Wuwu, Xiaoqi Li, Zhuoyang Liu, Ying Li, Renrui Zhang, Peng Jia, Pheng-Ann Heng, Shanghang Zhang

CVPR 2026 · Vision-language and multimodal › Robot action and navigation · approach: Vision-language-action foundation models

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 Manual CoT with planning expert (ManualVLA) vs hierarchical VLM + pi0.5ManualVLA · 2D LEGO Assemblysuccess rate: 0.60 -> 0.85Improved
ManualVLA vs hierarchical baselineManualVLA · 3D LEGO Assemblysuccess rate: 0.35 -> 0.65Improved
ManualVLA vs hierarchical baselineManualVLA · Object Rearrangementsuccess rate: 0.50 -> 0.65Improved
ManualVLA vs goal-conditioned CoT-VLAManualVLA · 2D LEGO Assemblysuccess rate: 0.30 -> 0.85Improved
add more manual content (text, subgoal image, position) to the manual (ablation)ManualVLA · 2D LEGO Assemblysuccess rateImproved
add explicit CoT (visual-prompt image) to implicit CoT (ablation)ManualVLA · 2D LEGO Assemblysuccess rateImproved
add implicit CoT (latent manual features) to explicit CoT (ablation)ManualVLA · 2D LEGO Assemblysuccess rateImproved
use MoT (duplicate full transformer) instead of MoE (duplicate FFN only) (ablation)ManualVLA · 2D LEGO Assemblysuccess rateImproved
use diffusion-based action generation (ablation)ManualVLA · 2D LEGO Assemblysuccess rateImproved
unseen backgroundManualVLA · 2D LEGO Assemblysuccess rate: 0.85 -> 0.65 (-23%)Hurt
unseen object shapeManualVLA · 2D LEGO Assemblysuccess rate: 0.85 -> 0.60 (-29%)Hurt
unseen lightingManualVLA · 2D LEGO Assemblysuccess rate: 0.85 -> 0.70 (-17%)Hurt

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.

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

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