DrivingDojo

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

• id: drivingdojo
• kind: world model and video
• robot or device: fleet vehicle cameras, Chinese cities (class: vehicle)
• how collected: fleet driving, curated
• size as stated: about 18k videos (~150 h)
• hours: 150 | hours only claimed: none | episodes: not stated
• year: 2024
• organisation: CASIA / Meituan
• licence: Apache-2.0 (class: open)
• access: gated (class: gated)
• link (dataset): https://huggingface.co/datasets/Yuqi1997/DrivingDojo
• Hugging Face repository: Yuqi1997/DrivingDojo, 486 downloads in 30 days, https://huggingface.co/datasets/Yuqi1997/DrivingDojo
• note: Size from the paper, not verified against it; card lists >1T category.

Confirm the licence at the link before relying on it.