Geometric Action Model for Robot Policy Learning

通用机器人策略既要遵循用户指令，也要推理物体、相机和机器人动作如何在三维物理世界中相互作用。近期视觉—语言—动作模型（VLA）和视频世界—动作模型（WAM）继承了大规模基础模型的强语义或时序先验，但仍主要工作在二维图像帧或二维派生潜空间中，因此接触密集操作所需的三维几何仍是隐式的。作者提出 Geometric Action Model（GAM），一种语言条件操作策略，直接把预训练几何基础模型（GFM）改造成感知、时序预测和动作解码的共享基底。GAM 在中间层切分 GFM：浅层作为观察编码器，在切分处插入因果未来预测器，并以语言、本体状态和动作历史为条件预测未来潜 token；这些未来 token 再经过剩余 GFM block 做特征传播与解码，使同一骨干同时生成未来几何和动作。该设计只做少量架构修改，就赋予 GFM 语言条件的时序世界建模能力，同时保留其丰富几何先验。作者在多组仿真和真实机器人操作基准上报告，GAM 相比当前基础模型规模基线更准确、更鲁棒、更快且更轻量。

Generalist robot policies must follow user instructions while reasoning about how objects, cameras, and robot actions interact in the 3D physical world. Recent vision-language-action models (VLAs) and video world-action models (WAMs) inherit strong semantic or temporal priors from large-scale foundation models, but they still operate primarily on 2D image frames or 2D-derived latent spaces, leaving implicit the 3D geometry required for contact-rich manipulation. We propose the Geometric Action Model (GAM), a language-conditioned manipulation policy that directly repurposes a pretrained geometric foundation model (GFM) as a shared substrate for perception, temporal prediction, and action decoding. GAM splits the GFM at an intermediate layer: the shallow layers serve as an observation encoder, and a causal future predictor inserted at the split layer forecasts future latent tokens conditioned on language, proprioception, and action history. The predicted future tokens are then routed through the remaining GFM blocks for feature propagation and decoding, allowing a single backbone to produce both future geometry and actions. This design equips the GFM with language-conditioned temporal world modeling through minimal architectural modification while preserving its rich geometric priors. Across a broad suite of simulation and real-robot manipulation benchmarks, GAM is more accurate, more robust, faster, and lighter than current foundation-model-scale baselines.