Official code release for:

Compose by Focus: Scene Graph-based Atomic Skills Paper: https://arxiv.org/abs/2509.16053

This repository implements an object-centric 3D diffusion policy pipeline that is effective for skill composition and long-horizon, multi-step robotic tasks.

You can try Compose by Focus directly in Google Colab without any local setup. It provides an easy-to-use implementation of our scene graph–based 3D diffusion policy.

Train:

python train.py

Evaluate (test suites):

python eval_test_together.py

Compose by Focus leverages scene graph-based object-centric representations to learn reusable atomic skills and compose them into complex behaviors. The policy is trained using a 3D diffusion model, enabling expressive and robust action generation under diverse scene configurations.

The compositional benchmark used in this work is available at: https://github.com/han20192019/skill-composition-benchmark

More detailed usage instructions is coming soon.

If you find this benchmark useful, please cite:

@article{qi25arxiv-compose,
  title={Compose by Focus: Scene Graph-based Atomic Skills},
  author={Qi, Han and Chen, Changhe and Yang, Heng},
  journal={arXiv preprint arXiv:2509.16053},
  year={2025},
  note={\linkToWeb{https://computationalrobotics.seas.harvard.edu/SkillComposition/}}
}