AI Paper Atlas

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

From Pixels to Tokens: A Systematic Study of Latent Action Supervision for Vision-Language-Action Models

Yihan Lin, Haoyang Li, Yang Li, Haitao Shen, Yihan Zhao, Chao Shao, Jing Zhang

ICML 2026 · Vision-language and multimodal › Robot action and navigation · approach: Vision-language-action foundation models · 0 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 image-based latent action supervision (LA-Direct)unified VLA baseline (Qwen3-VL-2B) · LIBERO-Longsuccess rate (%): 85.8 -> 96.6Improved
add action-based latent action tokens (LA-Tok)unified VLA baseline · RoboTwin 2.0 (4-task subset)avg success rate (%): 60.5 -> 78.0Improved
use image-based instead of action-based latent actions for long-horizon tasksVLA baseline · LIBERO-Longsuccess rate (%): +8.4% to +10.8% vs +6.8% (LA-Tok)Improved
explicit direct decoding (LA-Direct) instead of explicit conditional decoding (LA-Cond) (ablation)image-based latent actions · LIBERO-Longsuccess rate (%): 94.2 -> 96.6Improved
use LA-Cond instead of LA-Direct (ablation)image-based latent actions · RoboTwin 2.0 (4-task subset)avg success rate (%): 71.8 -> 73.8Improved
explicit supervision (LA-Direct) instead of implicit alignment (LA-Align) (ablation)image-based latent actions · LIBERO-Longsuccess rate (%): 94.8 -> 96.6Improved
discrete token supervision instead of continuous regression (ablation)LA-Direct · LIBERO (4-suite average)success rate (%): 94.4 -> 97.1Improved
discrete token supervision instead of continuous regression (ablation)LA-Tok · LIBERO (4-suite average)success rate (%): 93.3 -> 95.5Improved
align at early VLM layers instead of layer 17 (ablation)LA-Align · LIBERO (4-suite average)success rate (%): 95.0/97.0 -> 91.6Hurt
align at final layer instead of mid-to-late layer (ablation)LA-Align · LIBERO (4-suite average)success rate (%): 97.0 -> 95.0Hurt
lengthen action placeholders to match latent variants (ablation)unified VLA baseline · LIBEROsuccess rate (%): +0.9% (PH-Cond), -1.2% (PH-Direct)No clear effect
add latent action supervision (LA-Cond) under joint multi-task trainingunified VLA baseline · RoboTwin 2.0, 10 tasks joint trainingavg success rate (%): 53.60 -> 74.50Improved

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