用在线模仿预训练世界模型，提升机械臂少样本实机迁移效果
A Recipe for Efficient Sim-to-Real Transfer in Manipulation with Online Imitation-Pretrained World Models
- 结合在线模仿预训练与离线微调，利用仿真环境持续优化世界模型
- 在真实场景迁移中成功率提升23.3%，仿真到仿真提升31.7%
- 适合数据稀缺但需高鲁棒性机械臂操控的场景
我们关注在真实世界专家数据有限的情况下进行模仿学习。现有离线模仿方法常因数据覆盖不足导致性能严重下降。本文提出一种基于世界模型的在线模仿预训练框架，通过仿真环境中的在线交互缓解离线方法的数据覆盖缺陷，提升微调时的鲁棒性与泛化能力。实验表明，该方法在仿真到仿真的迁移中成功率提升至少31.7%，在仿真到真实的迁移中提升23.3%以上，显著优于现有离线模仿学习基线。
原文摘要 · Abstract (English)
We are interested in solving the problem of imitation learning with a limited amount of real-world expert data. Existing offline imitation methods often struggle with poor data coverage and severe performance degradation. We propose a solution that leverages robot simulators to achieve online imitation learning. Our sim-to-real framework is based on world models and combines online imitation pretraining with offline finetuning. By leveraging online interactions, our approach alleviates the data coverage limitations of offline methods, leading to improved robustness and reduced performance degradation during finetuning. It also enhances generalization during domain transfer. Our empirical results demonstrate its effectiveness, improving success rates by at least 31.7% in sim-to-sim transfer and 23.3% in sim-to-real transfer over existing offline imitation learning baselines.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。