Continuous Control of Diverse Skills in Quadruped Robots Without Complete Expert Datasets

Learning diverse skills for quadruped robots presents significant challenges, such as mastering complex transitions between different skills and handling tasks of varying difficulty. Existing imitation learning methods, while successful, rely on expensive datasets to reproduce expert behaviors. Inspired by introspective learning, we propose Progressive Adversarial Self-Imitation Skill Transition (PASIST), a novel method that eliminates the need for complete expert datasets. PASIST autonomously explores and selects high-quality trajectories based on predefined target poses instead of demonstrations, leveraging the Generative Adversarial Self-Imitation Learning (GASIL) framework. To further enhance learning, We develop a skill selection module to mitigate mode collapse by balancing the weights of skills with varying levels of difficulty. Through these methods, PASIST is able to reproduce skills corresponding to the target pose while achieving smooth and natural transitions between them. Evaluations on both simulation platforms and the Solo 8 robot confirm the effectiveness of PASIST, offering an efficient alternative to expert-driven learning.

@article{tu2025_2503.03476,
  title={ Continuous Control of Diverse Skills in Quadruped Robots Without Complete Expert Datasets },
  author={ Jiaxin Tu and Xiaoyi Wei and Yueqi Zhang and Taixian Hou and Xiaofei Gao and Zhiyan Dong and Peng Zhai and Lihua Zhang },
  journal={arXiv preprint arXiv:2503.03476},
  year={ 2025 }
}