Unverified50% confidenceFactExact time
Thinker-WM's training corpus totals approximately 20K hours of first-person video, 10K hours of real robot data, and 1K hours of simulation data, of which about 4K hours of real robot data has aligned depth supervision.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRiemann 1.0训练数据总规模达23.2万小时，覆盖人类第一视角视频、UMI或外骨骼手套数据及机器人轨迹数据76% similarUnverifiedUnitree has open-sourced the humanoid robot foundation model Unifor LMW-LA 1.0, with a parameter scale of 6 billion, trained on approximately 2,500 hours of real robot data; a single model can perform 64 tasks including getting out of bed, folding clothes, and sorting, and is compatible with multiple types of robotic hands.75% similarUnverifiedTeleoperation demonstration data has long been the mainstream data source for robot imitation learning, but each frame of data requires a real operation, collecting a multi-step task often takes tens of minutes, and it is difficult to scale.69% similarUnverifiedRobot teaching refers to the process in which an engineer holds a teach pendant and guides the robotic arm point by point to record trajectory points such as joint angles, speeds and wait times; a process with dozens of motions typically requires hundreds of teaching points and takes several hours to several days.68% similarUnverified人形机器人的感知-决策-执行全链路时延预算通常需控制在10-50毫秒以内才能保证动态稳定性，其中视觉处理和深度学习推理延迟最大（5-30ms）68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1024083API
curl https://kongchang.com/api/v1/knowledge/claims/1024083MCP
get_claim(id=1024083)