Skip to content

Robotics and CAD

BEHAVIOR / BEHAVIOR-1K

Long-horizon everyday household activities and object interactions.

About BEHAVIOR / BEHAVIOR-1K

What it measures

Task completion in the specified simulator/task suite.

Scope & limitations

Licensing, compute and embodiment requirements differ from text-agent evaluations; demonstration datasets and evaluation episodes must remain separate.

Evaluation environment

Compatible robot policy and supported simulation assets.

Source

Original benchmark