Repolode
Submit
StanfordVL

StanfordVL/BEHAVIOR-1K

A simulation benchmark for testing embodied AI agents on 1,000 everyday household activities such as cleaning, cooking, and organizing.

View on GitHubWebsite
Stars
1,796
Forks
262
Open issues
313
Last commit
yesterday

What is it?

BEHAVIOR-1K is a simulation benchmark for embodied AI agents, covering 1,000 everyday household activities. The activities were chosen from real human time-use surveys and preference studies, so the tasks reflect what people actually do at home, such as cleaning, cooking, and organizing.

The repository is a monolithic codebase that bundles what is needed to train and evaluate agents on these human-centered tasks. An installation script handles dependencies and lets users install only the components they need. Full setup instructions are provided in the official installation guide.

Researchers working on robot learning, household task planning, or embodied AI evaluation can use it as a shared benchmark. The project is published alongside an arXiv paper, which is cited in the repository.

Key features

  • 01Covers 1,000 everyday household activities drawn from human time-use surveys
  • 02Provides a modular installation script for selecting individual components
  • 03Supports training and evaluating embodied AI agents on human-centered tasks
  • 04Released under the MIT license

Use cases

  • Benchmarking embodied AI agents on realistic household chores
  • Training robot policies for cleaning, cooking, and organizing tasks
  • Comparing task-planning methods on a shared set of everyday activities

Quick start

  1. 1

    Run the installation script from the README (it is a script, so no command is shown)

  2. 2

    Follow the Installation Guide on the project website for modular setup

Things to know

  • The README gives only an installation script and a link to the guide, with no direct package-manager command.
  • Setup covers a full simulation stack, which may be heavy to install and run.

In collections

Reinforcement learning frameworks & training environmentsRobotics & embodied AI projects

Topics

#embodied-ai#robotics#simulation#benchmark#reinforcement-learning

FAQ

What is BEHAVIOR-1K?
A simulation benchmark for testing embodied AI agents on 1,000 everyday household activities such as cleaning, cooking, and organizing. BEHAVIOR-1K is a simulation benchmark for embodied AI agents, covering 1,000 everyday household activities. The activities were chosen from real human time-use surveys and preference studies, so the tasks reflect what people actually do at home, such as cleaning, cooking, and organizing.
Is BEHAVIOR-1K free to use? What license does it have?
BEHAVIOR-1K is open source under the MIT license. Check the license terms before using it in commercial products.
Is BEHAVIOR-1K still maintained?
Actively maintained. The last commit to BEHAVIOR-1K was on Nov 6, 2026.
What can I use BEHAVIOR-1K for?
Benchmarking embodied AI agents on realistic household chores Training robot policies for cleaning, cooking, and organizing tasks Comparing task-planning methods on a shared set of everyday activities
How do I install BEHAVIOR-1K?
Run the installation script from the README (it is a script, so no command is shown) Follow the Installation Guide on the project website for modular setup
Where can I get BEHAVIOR-1K?
The source code and installation instructions are on GitHub: https://github.com/StanfordVL/BEHAVIOR-1K

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects