自研高保真物理仿真/合成数据引擎,据称生成速度达实时的 43 万倍,为 RFM 提供训练数据
17自由度灵巧手,2025年7月与B06同代发布,此前CES 2025展出第一代17DOF灵巧手
All-scenario compute core for embodied AI; the S100P (128 TOPS) appeared in Vbot's consumer robot dog at CES 2026
Dedicated-scenario manipulation foundation model with task-level closed loops, supporting ultra-long-horizon multi-task continuous operation
VLA with open-world generalization; can control mobile manipulation robots to tidy kitchens and bedrooms never seen before
Small open-source humanoid robot for the developer ecosystem
Wheeled liftable humanoid design, 34+ DoF, 6-hour battery life, ±0.02mm assembly accuracy; the world's first general intelligent robot carrying the Alpha Brain embodied foundation model; achieved 100-unit real monthly delivery in Dec 2025
Co-developed with Huawei and China Mobile; the world's first 5G-A humanoid torchbearer (per original notes)
首代 VLA 模型,与 Gemini Robotics-ER 一同发布,将 Gemini 多模态能力引入机器人
全球首个开放人形机器人基础模型,快慢双系统架构,开源权重;后续迭代 N1.5 等版本
合成数据生成管线(基于 Omniverse/ Cosmos),帮助开发者以仿真批量生成模仿学习训练数据
通用具身智能软件平台,支持多本体、多场景协同的「大脑+小脑」一体化架构
全尺寸双足人形机器人:身高167cm、体重70kg、全身40个自由度(不含灵巧手)、最大行走速度1m/s、末端负载1.5kg;2026年4月13日京东开售,定价28.58万元
Open-source π0 weights and code plus the π0-FAST autoregressive model; supports PyTorch, with DROID dataset training recipes
The first video-generation foundation model algorithm for humanoid robots
1,000 sqm R&D space; 40 devices producing 6,000 data entries daily, with 200,000 structured interaction entries and 1,300 hours of continuous motion data accumulated
收购沃尔玛 Advanced Systems and Robotics 业务后推出的店内履约方案(承接原 Alert Innovation 技术)
机器人视觉相机,AC1「移动之眼」、AC2「操作之眼」,配AC Studio工具链