Dataset profile

BridgeData V2

BridgeData V2 supports open-vocabulary, multi-task robot learning conditioned on goal images or natural-language instructions. It is a general household-manipulation benchmark, not a healthcare dataset.

Source institutionUC Berkeley and collaborators
Robot ManipulationHousehold Robotics

Overview

BridgeData V2 supports open-vocabulary, multi-task robot learning conditioned on goal images or natural-language instructions. It is a general household-manipulation benchmark, not a healthcare dataset.

What's included

  • WidowX robot trajectories
  • Natural-language instructions per trajectory
  • RGB-D and multi-camera observations

Tasks / activities

pick and placepushingfolding clothopening doors and drawerssweeping granular media

Modalities

RGB-DRobot ActionsLanguageAnnotations

Collection methodology

Demonstrations were collected through robot teleoperation across varied environments and objects.

Potential applications

  • Language-conditioned manipulation
  • Offline robot learning
  • Generalization to novel objects
  • Goal-image conditioning

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 BridgeData V2 publisher terms.

Limitations / opportunities for custom data

Custom collection can bring healthcare tasks, professional human workflows, facility constraints, and care-specific annotations.

Need data beyond this benchmark?

We design custom healthcare and caregiving collections around the tasks, experts, environments, and modalities your model requires.

Request Similar Data