26 robotics datasets you can order online today, across 5 categories — from egocentric video and robot teleoperation to hand tracking and motion capture. Priced per recorded hour.
Priced per recorded hour, 10 hour minimum. Card payment through Stripe. After payment we confirm availability and arrange delivery with you by email. Every order is licensed under the RC Standard Data License v1.
Head-mounted RGB-D capture of tabletop and hand-over tasks with reconstructed 3D hand keypoints, head and hand 6-DoF pose, IMU, synchronized audio and camera calibration. Additional hours available on order.
Head-mounted 4K egocentric video of skilled manual trades (electronics repair, sewing, shoe repair, woodwork, food prep, retail tagging) in unedited 2–13 minute takes, with per-video segmentation manifests (task segments, idle periods, hand dominance). Additional hours available on order.
Household manipulation (laundry folding, dishes, surface cleaning, cooking) recorded with a head camera synchronized to seven upper-body IMUs at 24 Hz, with a human-to-robot arm retarget on every episode. Additional hours available on order.
Head-mounted 4K egocentric video of household chores in real homes (cooking, dishwashing, mopping, peeling fruit, making coffee, watering plants), with a per-clip sheet listing objects, surface type and clutter level. Additional hours available on order.
Egocentric 1080p video of household tasks (folding clothes, cooking, cleaning, dishes), delivered raw without task labels. Additional hours available on order.
Head-mounted egocentric video on small production lines (folding, food production, packaging, sewing, garment printing), with the task named per clip. Additional hours available on order.
The same household action recorded simultaneously from a head-mounted camera and two fixed third-person cameras, all in 4K, for ego-to-exo view-transfer research. Additional hours available on order.
Egocentric task videos in workstation and in-home scenes, each with a JSON annotation carrying three natural-language instruction variants, an object list and scene metadata. Additional hours available on order.
Egocentric 1080p video of full household workflows (kitchen cleaning, cutting vegetables, peeling fruit, folding clothes, tidying shelves, sweeping), with the task named per clip. Additional hours available on order.
Egocentric RGB with aligned depth and raw accelerometer/gyroscope at 268 Hz, plus a per-frame timestamp table for cross-sensor sync verification, suited to VIO and sensor-fusion pretraining. Additional hours available on order.
Egocentric 1080p video of five household tasks (vegetable cutting, laundry folding, dishwashing, table sorting, box sorting) with English task labels. Additional hours available on order.
Low-resolution (640×480) egocentric RGB clips of bare-hand activity, unlabelled. Additional hours available on order.
Side-by-side stereo egocentric video captured on a VR headset during desk, toy and cabinet tidying tasks, supporting 3D hand and scene reconstruction without a depth sensor. Additional hours available on order.
Teleoperated humanoid episodes on two robot configurations with multi-finger dexterous hands and three RGB cameras (head and both wrists), covering office and logistics tasks such as document sorting, room tidying and parcel weighing. Additional hours available on order.
Teleoperation recordings across eight single-arm and bimanual platforms in one ROS2 MCAP format, covering insertion, pick-and-place, t-shirt folding, tossing and separation, with 1–4 compressed RGB streams plus joint telemetry per recording. Additional hours available on order.
Dual-arm teleoperation with parallel grippers and three cameras (top and both wrists) on storage and organization tasks, with per-episode labels, in native LeRobot v3 at 30 fps. Additional hours available on order.
Teleoperated household and sorting episodes with head and two wrist cameras, headset and wrist 6-DoF pose at 30 Hz, and millisecond-level action segment labels. Additional hours available on order.
Teleoperated 5-DoF soft continuum arm with gripper performing hand-over grasps, recorded from base and wrist cameras. Additional hours available on order.
Human hand demonstrations captured by 14 synchronized cameras (11 RGB + 3 RGB-D) with a high-density tactile glove, delivered with a retarget to a dexterous robot hand. Additional hours available on order.
Bimanual household manipulation (bed making, pillow arrangement) with 75-D joint angles per hand at ~50 Hz, six calibrated fisheye cameras and chest-mounted point clouds at ~10 Hz, with a task description and success flag per action. Additional hours available on order.
Full hand skeleton (per-joint position and rotation) from a hand-tracking glove at 120 Hz, covering object grasps and free gestures; hand-only, no video. Additional hours available on order.
Head-mounted six-camera egocentric recordings with VIO, plus handheld gripper sessions with end-effector pose, IMU and encoder telemetry; ships ROS packages (URDF, STL meshes, Gazebo launch) for both devices. Additional hours available on order.
Paired left/right handheld gripper recordings with 6-DoF pose at 100 Hz, gripper opening width at 200 Hz and wrist RGB at 60 fps. Additional hours available on order.
Full-body skeleton from an inertial motion-capture suit at 90 Hz, covering locomotion and exercise (jumping, kneeling, running, workouts) in long continuous takes. Additional hours available on order.
Marker-based optical full-body skeleton at 120 Hz, covering dance, squats, crawling and other posture changes. Additional hours available on order.
Tabletop manipulation captured with 59-bone body and finger tracking, 129-point hand pressure, and head-mounted plus two fixed RGB-D cameras with full calibration. Additional hours available on order.
The marketplace connects data collectors with ML teams who need high-quality training data.
Use your own robot or rent from Robotics Center. Record teleoperation demonstrations with our data platform.
Upload datasets in HDF5, RLDS, or LeRobot format. Add metadata: robot type, task, camera config.
Choose your license (commercial, research-only, exclusive). Set your price per episode or dataset.
Search by robot type, task category, format, and quality score. Compare datasets side by side.
Watch video replays of sample episodes. Inspect action distributions and camera views before buying.
We confirm availability, then deliver by secure transfer. Train on your own infrastructure or on the RC platform.
Turn your robot demonstrations into revenue. List your datasets on the marketplace and reach thousands of ML researchers worldwide.
