We’re powering an era of physical agents — enabling robots to perceive, plan, think, use tools and act to better solve complex, multi-step tasks.

NEURA Robotics在杭州新设中心，推动物理AI与机器人训练，构建全球数据生态“Neuraverse”。该中心将促进本地人才培养与国际合作，加速人形机器人商业化进程。

我们推出Gemini Robotics 1.5和Gemini Robotics-ER 1.5模型，旨在提升机器人理解和执行复杂任务的能力。前者将视觉信息转化为动作指令，后者支持多步骤计划和数字工具调用。这些进展将使机器人更高效地完成多步骤任务。

The field of robotics presents opportunities in industries from manufacturing to healthcare to agriculture. Researchers are exploring the ethical and safety implications to set safe guardrails.

我们推出了高效的本地机器人模型Gemini Robotics On-Device，具备灵活性和快速适应能力，独立于网络运行，适合延迟敏感应用，确保在不稳定环境中的稳健性。开发者可通过SDK轻松评估和调整该模型。

We’re introducing an efficient, on-device robotics model with general-purpose dexterity and fast task adaptation.

We’re introducing an efficient, on-device robotics model with general-purpose dexterity and fast task adaptation.

We’re introducing an efficient, on-device robotics model with general-purpose dexterity and fast task adaptation.

Today, Google DeepMind announced a new family of Gemini models designed for robotics. Gemini Robotics is a vision-language-action (VLA) model that takes natural language…

Google DeepMind is announcing two new models, based on Gemini 2.0, that are designed for robotics.Gemini Robotics is our most advanced vision-language-action model, whic…

谷歌DeepMind推出Gemini Robotics，结合AI与物理世界，提升机器人在复杂任务中的表现。新模型具备通用性、互动性和灵活性，能够适应不同环境并执行精细操作，同时增强空间理解，提高机器人安全性与智能。

π0: A Foundation Model for Robotics with Sergey Levine - 719

本文探讨了RLAIF微调VLA模型的研究，强调通过偏好对齐提升机器人策略的泛化能力。当前具身模型的发展趋势与大语言模型相似，采用预训练-微调模式，模型参数规模不断扩大，逐步实现“一模型搞定所有任务”的目标。

本研究提出RDMM框架，以解决机器人自主决策能力不足的问题。通过实时信息处理和上下文感知，实验结果显示其规划准确率达到93%，为人机交互和机器人自主性提升提供了新机遇。