RoboVerseOrg/RoboVerse

RoboVerse: Towards a Unified Platform, Dataset and Benchmark for Scalable and Generalizable Robot Learning

What it solves

RoboVerse addresses the fragmentation in robot learning by providing a unified platform, dataset, and benchmark. It aims to enable scalable and generalizable robot learning by consolidating various tasks, robots, scenes, and assets into a single ecosystem.

How it works

RoboVerse operates as a downstream package (roboverse-py) that integrates with a core simulation engine called metasim. It leverages a wide array of existing simulation frameworks (such as Isaac Lab, MuJoCo, SAPIEN, and PyBullet) and integrates data from numerous robot learning projects (including RLBench, ManiSkill, and Meta-World) to provide a comprehensive environment for training and testing agents.

Who it’s for

This project is designed for researchers and developers working on embodied AI and robot learning who need a standardized way to evaluate generalizability and scalability across different robotic tasks and environments.

Highlights

  • Unified platform for scalable and generalizable robot learning.
  • Integration of multiple simulation frameworks including Isaac Lab, MuJoCo, and SAPIEN.
  • Aggregation of data and tasks from a vast number of existing robot learning benchmarks and datasets.
  • Support for a diverse range of robots, scenes, and assets.

Related

  • Project
    vikashplus/robohive

    A unified framework for robot learning that provides a collection of MuJoCo-simulated environments and tasks exposed via the OpenAI-Gym API.

  • Project
    ARISE-Initiative/robosuite

    A MuJoCo-powered simulation framework and benchmark suite for robot learning, providing modular tools to create environments and test AI control algorithms.

  • Project
    RoboDojo-Benchmark/RoboDojo

    A unified sim-and-real benchmark for evaluating generalist robot manipulation policies across 60 tasks and five capability dimensions.

  • Project
    ARISE-Initiative/robomimic

    A framework for robot learning from demonstration that provides standardized datasets and offline learning algorithms to make robot manipulation research accessible and reproducible.

  • Project
    galilai-group/stable-worldmodel

    A platform for reproducible world model research and evaluation that provides a unified interface for data collection, training, and model-predictive control evaluation.