Vision-Language-Action (VLA) models are fundamentally bottlenecked by the scarcity of expert demonstrations -- triplets of observations, instructions, and actions that are costly to collect at scale. We argue that this bottleneck stems from conflating two distinct learning objectives: acquiring physical competence (how to move) and acquiring semantic alignment (what to do). Crucially, only the latter requires language supervision. Building on this Decomposition Hypothesis, we propose Task-Agnostic Pretraining (TAP), a two-stage framework that first learns transferable motor priors from cheap, unlabeled interaction data -- including discarded off-task trajectories and autonomous robot play -- via a self-supervised Inverse Dynamics objective. A lightweight second stage then grounds these priors in language using minimal expert data. On the SIMPLER benchmark, TAP matches models trained on over 1M expert trajectories while using orders of magnitude less labeled data, yielding a 10% absolute gain over standard behavior cloning. On a real-world WidowX platform, TAP retains 25% success under camera perturbations where internet-scale baselines collapse to 0%, demonstrating that task-agnostic pretraining produces robust, transferable physical representations and offers a scalable path forward for Embodied AI.
Microsoft Research Asia（微软亚洲研究院）；Nanjing University（南京大学）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Wuhan University（武汉大学）；University of Technology Sydney（悉尼科技大学）；Tsinghua University（清华大学）
What Matters in Quality-Aware VLA Pretraining: A Benchmark and Empirical Study
Georgia Institute of Technology（佐治亚理工学院）；Axis Robotics（Axis机器人公司）；University of California, Riverside（加州大学河滨分校）；Carnegie Mellon University（卡内基梅隆大学）；University of California, Berkeley（加州大学伯克利分校）；Texas A&M University（德克萨斯农工大学）；National University of Singapore（新加坡国立大学）；Rice University（莱斯大学）；Nanyang Technological University（南洋理工大学）
PixelVLA: Advancing Pixel-level Understanding in Vision-Language-Action Model
University of Trento（特伦托大学）；School of Automation Science and Engineering, South China University of Technology（华南理工大学自动化科学与工程学院）；Mohamed bin Zayed University of Artificial Intelligence（马尔代夫人工智能大学）；Australian National University（澳大利亚国立大学）
MVP-LAM: Learning Action-Centric Latent Action via Cross-Viewpoint Reconstruction
Seoul National University, Seoul, South Korea（首尔国立大学，首尔，韩国）；Konkuk University, Seoul, South Korea（韩国konkuk大学，首尔，韩国）；Microsoft Research Asia, Beijing, China（微软亚洲研究院，北京，中国）；HodooAI Labs, Seoul, South Korea（HodooAI实验室，首尔，韩国）
Chain of World: World Model Thinking in Latent Motion
Harbin Institute of Technology（哈尔滨工业大学）；Li Auto；Beijing Academy of Artificial Intelligence (BAAI)（北京人工智能研究院）；University of New South Wales（新南威尔士大学）；Chongqing Research Institute of HIT（哈尔滨工业大学重庆研究院）；Peking University（北京大学）
