RDT2 UMI dataset

From the catalogue. Listed by one researcher from one source page.

• id: rdt2-umi-dataset
• kind: handheld gripper
• robot or device: redesigned bimanual UMI gripper with HTC Vive tracking (class: other)
• how collected: handheld gripper, not teleop
• size as stated: 10,000+ hours, 100+ scenes
• hours: not stated | hours only claimed: 10,000 | episodes: not stated
• year: 2025
• organisation: RDT team (thu-ml)
• licence: not stated (class: not stated)
• access: not released when checked; page says data will be released shortly (class: not stated)
• link (dataset): https://rdt-robotics.github.io/rdt2/
• note: Largest claimed UMI corpus; only models and example data are public.

Confirm the licence at the link before relying on it.