Dataset profile
DROID
DROID is a large robot manipulation dataset designed to improve generalization across environments, tasks, and collectors. It is relevant as a comparison point for teams deciding where human healthcare demonstrations need to complement robot-native data.
Overview
DROID is a large robot manipulation dataset designed to improve generalization across environments, tasks, and collectors. It is relevant as a comparison point for teams deciding where human healthcare demonstrations need to complement robot-native data.
What's included
- Robot manipulation trajectories
- Interaction data across many scenes and tasks
- Camera, robot state, action, and language annotation resources
Tasks / activities
Modalities
Collection methodology
Robot data was collected across distributed scenes and tasks by data collectors using a shared robot manipulation platform.
Potential applications
- Robot manipulation pretraining
- Policy generalization
- Language-conditioned control
- Embodied AI benchmarking
Access & licensing
This is a third-party public/research dataset. We do not own or distribute this dataset. Access and usage are governed by the original publisher's terms.
Access and use are governed by the original DROID publisher terms.
Limitations / opportunities for custom data
Healthcare buyers may need domain-expert human demonstrations, care settings, and workflow labels that are not present in general robot manipulation data.
Need data beyond this benchmark?