LIBERO
Lifelong roBotic lEarning benchmaRk. 130 tasks, 65K demos. Emergent Mind.
Overview
LIBERO provides a benchmark for lifelong robot learning with multiple task suites (spatial, object, goal) and 65K expert demonstrations. Built on RoboSuite simulation.
Scale & Content
- 130 tasks
- 65,000 demonstrations
- Spatial, object, goal suites
- RoboSuite (MuJoCo) simulation
Official Links
- libero-project.github.io/datasets — Datasets & benchmark
Citation
See the project site for BibTeX and paper references.