Three core types of robot training data
Egocentric
Head-, chest- or glasses-mounted first-person data with RGB, depth, IMU and hand trajectories — a real human perspective for vision-language-action models.
UMI Hand-held
Embodiment-free human demonstrations captured with hand-held devices: 6D poses, end-effector trajectories and grasp states, for low-cost skill collection at scale.
Teleoperation
Multi-camera video plus full qpos/qvel/action signals from real-robot teleoperation — high-value data fully aligned with the deployment embodiment.
Hand-picked, high quality
单臂抓取真机遥操作数据集
UMI 双手协作多源对齐数据集
双臂协作整理真机遥操作数据集
Ego 桌面整理双目第一视角数据集
UMI 手持桌面抓取数据集（Quest 同步）
仓储分拣真机遥操作数据集
A closed loop from capture to delivery
Capture
Ego / UMI / teleoperation multi-source capture
Governance
Anonymization, quality scoring and label checks
Preview
Real MCAP video, trajectory and schema previews
Trade
Fixed price, quote or subscription models
Delivery
License management with full download audit logs
Search by application scene
Recently listed
test
UMI 手持桌面抓取数据集（Quest 同步）
Ego 桌面整理双目第一视角数据集
Join the StellarRobotics data ecosystem
Whether you supply data or train models, publish, discover and deliver high-quality robot data here.