Feagine Ego is a lightweight first-person 6-lenses data acquisition headband for embodied intelligence, robot imitation learning, and real-world operation data acquisition.

Model generalization depends on precise data alignment. Feagine Ego uses six-camera panoramic vision and IMU sensing,
ensuring temporal consistency between visual and motion trajectories.

The overall weight is 300 g, balancing comfort and stability during long-term wear, suitable for scenarios such as Data Acquisition Data Factory, robot teleoperation, imitation learning, and embodied AI data construction.
Six-channel visual data and IMU data are collected using a unified timestamp, supporting hardware-level time synchronization to ensure consistency of MultiModal Machine Learning data during training, alignment, and playback.
Complete Calibration of Intrinsic and Extrinsic Parameters of camera, facilitating subsequent multi-view fusion, spatial reconstruction, trajectory alignment, and data processing for robot imitation learning.
Feagine Ego adopts a local hard storage solution, supporting single-machine local single-save of 3 hours of continuous operation data. After data acquisition is completed, data can be directly copied via an external USB drive, without the need for an online cloud storage platform or dedicated network cables, and there are no additional requirements for the network at the data factory site.

Calibration is performed using a dedicated calibration rig. The reprojection RMS of the three stereo pairs is 0.5529 px, 0.6837 px, and 0.6328 px respectively, while the six monocular calibrations achieve a reprojection RMS of 0.161–0.288 px. The camera intrinsics, distortion parameters, and stereo extrinsics show strong overall geometric consistency.
Bodyless Embodied Data →