Contributing

This page summarizes how you can contribute to FGManip and related open-source efforts for fine-grained manipulation research.

Summary

Contributing typically involves:

• Objects and parts: Add or extend objects in the part-annotated object library, including part annotations, grasp poses, and affordances for fine-grained manipulation.

• Skills: Contribute reusable skills to the skill library that can be composed into new tasks.

• Tasks and demonstrations: Propose new tasks, define task logic and reward structures, and provide demonstrations. This includes proper typing and labelling (e.g., supported reward modes, supported agents), and registering tasks in the platform.

• Robots and assets: Add or adapt robot models and scene assets that integrate with the simulation environment.

We welcome bug reports, feature requests, benchmark suggestions, and code contributions. Please share issues and feedback through the project repository issue tracker and discussions.