Available Datasets

RoboX provides structured egocentric datasets designed for robotics imitation learning and perception.

All datasets are:

• First-person (egocentric)

• Campaign-scoped and opt-in

• Anonymized on-device

Campaign	Sensors	Description	Use Cases
EgoDaily	Camera + IMU	First-person recordings of daily activities captured during natural movement	Humanoid imitation learning, activity recognition, scene understanding
EgoGrasp	Camera + IMU + LiDAR on supported devices	Close-range hand-object interaction data	Dexterous manipulation models, grasp prediction, object affordance learning
EgoNav	Camera + IMU + Wi-Fi/BLE Signals	Indoor navigation traces through malls, offices, airports, staircases, and dense public environments	GPS-denied navigation, visual SLAM, obstacle avoidance
EgoScene	Camera + IMU + LiDAR on supported devices	360° room rotations and spatial scans for layout reconstruction	3D mapping, AR-assisted robotics, indoor spatial modeling
EgoSocial	Camera + IMU + Optional Audio With Consent	Recorded human interactions including gestures, conversations, and collaborative actions	Human-robot interaction, social robotics, multi-person activity recognition

EgoSocial is coming soon. All other campaigns are active for private beta testing.

More Datasets Coming

Expanded task domains, complex multi-person interaction data, advanced scene reconstruction, and head-mounted capture inputs.