DemoGen: Synthetic Demonstration Generation for Data-Efficient Visuomotor Policy Learning

Zhengrong Xue, Shuying Deng, Zhenyang Chen, Yixuan Wang, Zhecheng Yuan, Huazhe Xu

Track: system paper (must be submitted with a supplementary video)

Keywords: imitation learning, data generation, robotic manipulation

Abstract: Visuomotor policies have shown great promise in robotic manipulation but often require substantial amounts of human-collected data for effective performance. A key reason underlying the data demands is their limited spatial generalization capability, which necessitates extensive data collection across different object configurations. In this work, we present *DemoGen*, a low-cost, fully synthetic approach for automatic demonstration generation. Using only one human-collected demonstration per task, *DemoGen* generates spatially augmented demonstrations by adapting the demonstrated action trajectory to novel object configurations. Visual observations are synthesized by leveraging 3D point clouds as the modality and rearranging the subjects in the scene via 3D editing. Empirically, *DemoGen* significantly enhances policy performance across a diverse range of real-world manipulation tasks, showing its applicability even in challenging scenarios involving deformable objects, dexterous hand end-effectors, and bimanual platforms. Furthermore, *DemoGen* can be extended to enable additional out-of-distribution capabilities, including disturbance resistance and obstacle avoidance.

Anonymization: This submission has been anonymized for double-blind review via the removal of identifying information such as names, affiliations, and identifying URLs.

Presenter: ~Zhengrong_Xue1

Format: Yes, the presenting author will attend in person if this work is accepted to the workshop.

Funding: Yes, the presenting author of this submission falls under ICLR’s funding aims, and funding availability would significantly influence their ability to attend the workshop in person.

Submission Number: 22