Order from Lab 813633

Robot teleoperation

Real-robot teleoperation episodes: 2 to 4 synchronized camera videos, joint states, the operator's actions and a success label, all on one clock, with the hardware described. About 2,000 hours across 50+ tasks and 20+ sites, with failed attempts kept, licensed for commercial training with consent of everyone on camera.

How answering works

Any number of people fill one order. The lab receives one dataset, in its own columns.

Your agent reads the order
It looks at what you hold and says which columns you can fill, and what is missing.
Mora checks what you send
Every record: no copies, no personal data, not already public, not on Mora from someone else.
The lab pays per record it accepts
The lab named the price. You fill as much of the order as you can.

The columns the lab wants

You do not need every column. Mora says which ones each dataset fills.

Column Type What it must hold

episode_id, task_id, site_id, robot_id, operator_id id Which attempt, task, place, robot and person, as generated ids

instruction text The task in words, e.g. put the mug in the rack

video mp4 per camera, 30 fps or more, 480p or Wrist and fixed cameras, with a timestamp per frame, intrinsics and extrinsics

joint_position, joint_velocity, ee_pose, gripper_position float arrays at 15 Hz or more Robot state on the same clock as the video

action float array What the operator commanded at each step, with the action space stated (joint target or end-effector delta)

success, failure_reason, judged_by label Whether the task was done, why not, and who judged it

arm_model, gripper_model, urdf, teleop_device, control_hz text The hardware and control rate, so episodes can be retargeted to our arms

collection_mode label teleoperated, scripted or policy rollout, per episode

Column	Type	What it must hold
`episode_id, task_id, site_id, robot_id, operator_id`	id	Which attempt, task, place, robot and person, as generated ids
`instruction`	text	The task in words, e.g. put the mug in the rack
`video`	mp4 per camera, 30 fps or more, 480p or	Wrist and fixed cameras, with a timestamp per frame, intrinsics and extrinsics
`joint_position, joint_velocity, ee_pose, gripper_position`	float arrays at 15 Hz or more	Robot state on the same clock as the video
`action`	float array	What the operator commanded at each step, with the action space stated (joint target or end-effector delta)
`success, failure_reason, judged_by`	label	Whether the task was done, why not, and who judged it
`arm_model, gripper_model, urdf, teleop_device, control_hz`	text	The hardware and control rate, so episodes can be retargeted to our arms
`collection_mode`	label	teleoperated, scripted or policy rollout, per episode

How much, in the lab's words

2,000 hours, about 100,000 episodes, 50+ tasks, 20+ sites, 500+ objects, 10 to 20% failed attempts kept and marked. Roughly 5 TB. Counted in active (moving) hours and episodes, not raw hours.

episodes
video
state
actions
outcome
tasks
embodiment
extra-sensors