robocasa/robocasa

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

1677 stars · 243 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

RoboCasa is a large-scale simulation framework for training and benchmarking generally capable robots on everyday tasks, focused on kitchen environments. RoboCasa365 provides 365 tasks, 2,500+ kitchen scenes, thousands of 3D assets, 2,200+ hours of demonstration data, and benchmarking support for policy learning methods like Diffusion Policy, pi0, and GR00T.

Use cases

• train robot policies on simulated everyday kitchen tasks
• benchmark imitation learning methods like diffusion policy
• generate synthetic robot demonstration trajectories
• evaluate generalist robot foundation models in sim
• collect human demonstrations via teleoperation
• hierarchical policy learning with subtask-annotated datasets

When to choose

• you need large-scale simulated environments for robot manipulation research
• you want to benchmark policy learning methods like Diffusion Policy or GR00T
• you need demonstration datasets for training generalist robot policies
• you are researching multi-task or lifelong learning for robots

When to avoid

• you need real-world robot hardware control
• you need non-kitchen or industrial simulation environments
• you want a lightweight simulator without large asset downloads

Facets

library · maturity active

simulation machine-learning benchmarking data-generation robotics machine-learning simulation artificial-intelligence python cross-platform windows robot-learning physics-simulation humanoid-robot imitation-learning kitchen-tasks demonstration-datasets policy-learning robosuite linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="robocasa/robocasa")