Open Datasets Overview

This section mainly collects:

• Open-source datasets for embodied AI
• Data collected via imitation learning / teleoperation
• Data format conversion and normalization
• Data visualization and quality-check methods

You can create a dedicated document per dataset later, for example:

• oxe.md
• libero.md
• droid.md
• bridgev2.md

Each dataset article is recommended to cover:

• Source of the dataset and the range of tasks it covers
• Data format and field descriptions
• License
• Loading and preprocessing workflow
• Which models it is suitable for reproducing
• Common pitfalls encountered in practice