RoboSynChallenge: Mastering Real-World Dexterity via Generalizing Synthesized Manipulation Skills

👉Webpage | Document | Paper | Leaderboard

---

Contents

• Contents
• Installtion
• Datasets
• Training and Evaluation
• LeaderBoard

Installtion

Based on the EmbodiChain, we offer both docker and local installation methods. For detailed installation instructions, please refer to Installation Document.

Datasets

We provide 1,000 pre-collected trajectories per task as part of the open-source release RoboSynChallenge Dataset. The datasets hosted on HuggingFace are available at here.

However, we still strongly recommend users to perform data collection themselves. For detailed data collection instructions, please refer to Data Collection Document.

Training and Evaluation

Currently, RoboSynChallenge integrates training and evaluation for PI0, PI0.5, and Motus. Detailed procedures can be found in the documentation for the corresponding strategies: 👉Webpage. In addition, you can easily extend your own policys for training and evaluation by following the documentation 👉Webpage.

LeaderBoard

The full leaderboard and setting can be found in: https://edem-ai.github.io/robosynchallenge.github.io/#/leaderboard.