Google DeepMind 发布 Gemini Robotics 2.5 视觉-语言-动作模型
Google DeepMind 发布 Gemini Robotics 2.5,称其为目前能力最强的视觉-语言-动作(VLA)模型,在200项任务上相比此前基线成功率提升30%至100%,完成任务速度快20%至30%。
推荐理由:原文给出200项任务评测、Mercedes-Benz产线96%成功率与安全指标,可判断VLA模型在真实场景的进展。
ScaleVLA: Scaling Vision-Language-Action Models to 552 Billion Parameters
Ruixi Yu · Dairu Liu · Xuchuan Chen · Zekun Qi · Zekai Li · Zhuo Cao · et al.
前沿视觉语言模型(VLM)如GPT-6 Astra在机器人控制中展现出显著能力,表明广泛的预训练能提供对物理动作有价值的知识。然而,大多数视觉-语言-动作(VLA)策略仍依赖较小的VLM骨干,这一潜力尚未被充分利用。本文探究前沿规模的表示能否为机器人学习提供更强的基础。我们提出ScaleVLA,将552B参数的DeepSeek-V4.1-Flash骨干与可训练的动作DiT连接,并在统一训练设置下系统比较2B至552B的VLM。我们评估机器人基础模型的三个基本能力:在样本效率上,仅10次演示即超越用50次演示训练的2B骨干(87.8% vs. 85.6%);在泛化性上,LIBERO-PRO成功率高达45.1%,而较小骨干最多31.3%;在精确操作上,真实世界充电器插入成功率从4/10提升至9/10,增益集中在接触敏感的对齐环节。尽管规模庞大,ScaleVLA每个动作块的中位延迟为216毫秒。研究结果表明,前沿VLM不仅能作为推理代理,还能作为高效且可泛化机器人策略的强大表示基础。
尚无解读。
arXiv:2611.21607v1 Announce Type: new Abstract: Frontier Vision-Language Models (VLMs) such as GPT-6 Astra have demonstrated remarkable capabilities in robot control, suggesting that broad pretraining provides knowledge valuable for physical action. Yet most Vision-Language-Action (VLA) policies still rely on much smaller VLM backbones, leaving this potential largely untapped. We ask whether frontier-scale representations can provide a stronger foundation for robot learning. We introduce ScaleVLA, which connects a 552B DeepSeek-V4.1-Flash backbone to a trainable action DiT, and systematically compare VLMs from 2B to 552B under a shared training setup. We evaluate three fundamental capabilities of robot foundation models. In sample efficiency, just 10 demonstrations outperform a 2B backbone trained with 50 (87.8% vs. 85.6%). In generalization, LIBERO-PRO success reaches 45.1%, compared with at most 31.3% for smaller backbones. In precise manipulation, real-world charger insertion improves from 4/10 to 9/10 successes, with gains concentrated in contact-sensitive alignment. Despite its scale, ScaleVLA achieves 216 ms median latency per action chunk. Our findings suggest that frontier VLMs can serve not only as reasoning agents, but also as powerful representational foundations for efficient and generalizable robot policies.
Google DeepMind 发布 Gemini Robotics 2.5,称其为目前能力最强的视觉-语言-动作(VLA)模型,在200项任务上相比此前基线成功率提升30%至100%,完成任务速度快20%至30%。
推荐理由:原文给出200项任务评测、Mercedes-Benz产线96%成功率与安全指标,可判断VLA模型在真实场景的进展。
Google DeepMind 发布 Gemini Robotics ER 2,这是其最强大的具身推理模型。该模型可调用 Google Search 等工具,支持多步骤任务自纠错,并将执行交接给低层 VLA 模型。
推荐理由:原文给出了视频理解、任务编排和多机器人协作的具体性能数据,读者可以据此了解具身智能的最新能力边界和可用入口。
Google DeepMind 发布 Gemini Robotics 2,推出了三个模型:高级 VLA 模型可控制完整人形机器人从脚到指尖的运动,实现灵巧操作如系扣子;ER 2 推理模型可处理多步骤任务并协调多机器人协作;On-Device 2 模型可在几小时内适应全新机器人形态。
推荐理由:原文给出了三个模型的具体能力描述和可用入口,读者可以据此判断 Gemini Robotics 2 与前代及竞品的具体差异。