英伟达发布机器人基础模型GR00T 2,模型与数据将陆续开源
英伟达在CoRL大会上发布新一代世界动作模型GR00T 2,可同时预测未来画面与机器人动作,模型、数据、仿真环境和数据采集硬件将陆续开源,10万小时人类数据已率先开放。
推荐理由:英伟达GR00T 2的跨本体迁移路径与数据金字塔构成,可供读者判断机器人基础模型的训练范式变化。
Evo-World: A Lightweight Vision-Language-Action Model with Future Motion Prediction via Optical-Flow Supervision
Zile Zhong · Yuxin Du · Tao Lin · Haoxiao Wang · Wei Dai · Binhua Zuo · et al.
视觉-语言-动作( vision-language-action,VLA)模型为视觉理解、语言对齐与机器人控制提供了统一框架,但若能预测当前观测之外的场景演化,可为控制提供额外信息。仅依赖动作监督无法显式鼓励这种预测能力,而未来图像预测则会把任务相关运动与大量静态外观纠缠在一起。为克服上述局限,本文提出 Evo-World,一种轻量级 VLA 模型,通过光流( optical-flow)监督增强策略学习,提供聚焦于运动而非外观的任务相关预测信号。本文进一步设计训练流程,在策略学习过程中保留这些预测表征,在解耦表征学习与动作优化的同时调节表征多样性。在多个仿真基准上,Evo-World 表现优异,在 Meta-World MT50 上达到 91.1% 的层级平均成功率,相对非预测基线提升 13.0%。在真实场景 AgileX 机器人上,Evo-World 在单臂与双臂操作任务中均优于所评估的基线,同时仅使用 0.87B 参数,显存占用更低,推理效率更高。结果表明,基于光流监督的未来运动预测是一种有效且轻量的增强 VLA 模型预测能力的方法。
尚无解读。
arXiv:2610.12936v1 Announce Type: new Abstract: Vision-language-action (VLA) models provide a unified framework for visual understanding, language grounding, and robot control, yet anticipating how the scene will evolve beyond the current observation can provide additional information for control. Action supervision alone does not explicitly encourage such anticipation, while future-image prediction entangles task-relevant motion with largely static appearance. To address these limitations, we introduce Evo-World, a lightweight VLA model that augments policy learning with future motion prediction through optical-flow supervision, thereby providing a task-relevant predictive signal focused on motion rather than appearance. We further design the training procedure to preserve these predictive representations during policy learning, regulating representation diversity while decoupling representation learning from action optimization. Across simulation benchmarks, Evo-World demonstrates strong performance, achieving a tier-averaged success rate of 91.1% on Meta-World MT50 , with a 13.0% relative improvement over the non-predictive baseline. On real-world AgileX robots, Evo-World outperforms the evaluated baselines across single-arm and bimanual manipulation tasks while using only 0.87B parameters and delivering lower GPU memory usage and higher inference efficiency. These results demonstrate that future motion prediction through optical-flow supervision provides an effective and lightweight approach to enhancing the predictive capabilities of VLA models.
英伟达在CoRL大会上发布新一代世界动作模型GR00T 2,可同时预测未来画面与机器人动作,模型、数据、仿真环境和数据采集硬件将陆续开源,10万小时人类数据已率先开放。
推荐理由:英伟达GR00T 2的跨本体迁移路径与数据金字塔构成,可供读者判断机器人基础模型的训练范式变化。