DemoHLM is a framework for humanoid loco-manipulation that enables scalable data generation and policy learning across diverse tasks. For each task, DemoHLM requires only a single demonstration and automatically synthesizes hundreds to thousands of successful trajectories that complete the same task in varied environments.

Citation

If you find our work useful, please consider citing us!

@article{demohlm,
  title={DemoHLM: From One Demonstration to Generalizable Humanoid Loco-Manipulation},
  author={Fu, Yuhui and Xie, Feiyang and Xu, Chaoyi and Xiong, Jing and Yuan, Haoqi and Lu, Zongqing},
  journal={arXiv preprint arXiv:2510.11258},
  year={2025}
}