Scholarlink

Learning from demonstration (LfD), inspired by neuroscience, is an effective way for robotics to learn manipulation skills such as opening doors or grasping cups from human’s natural actions (Billard et al., in Springer handbook of robotics, 2008 [1]). Since the 1980s, plenty of methods have been proposed, e.g., Tanwani and Calinon (IEEE Rob Autom Lett 1(1):235–242, 2016 [2]), Khansari-Zadeh and Billard (IEEE Trans Rob 27(5):943–957, 2011 [3]) and Calinon et al. (IEEE Rob Autom Mag 17(2):44–54, 2009 [4]) used the Gaussian mixture model (GMM) and Gaussian mixture regression (GMR) to regenerate motions. Ng and Russell (Algorithms for inverse reinforcement learning, 2000 [5]) proposed the inverse reinforcement learning (IRL) method to build up an unknown reward function based on the observed trajectories to characterize solutions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Constrained DMP Framework for Robot Skills Learning and Generalization from Human Demonstrations

  • 章节

  • 发表日期：

    2025年

查看全文
  • Chenguang Yang,
  • Zhenyu Lu,
  • Ning Wang

摘要

Learning from demonstration (LfD), inspired by neuroscience, is an effective way for robotics to learn manipulation skills such as opening doors or grasping cups from human’s natural actions (Billard et al., in Springer handbook of robotics, 2008 [1]). Since the 1980s, plenty of methods have been proposed, e.g., Tanwani and Calinon (IEEE Rob Autom Lett 1(1):235–242, 2016 [2]), Khansari-Zadeh and Billard (IEEE Trans Rob 27(5):943–957, 2011 [3]) and Calinon et al. (IEEE Rob Autom Mag 17(2):44–54, 2009 [4]) used the Gaussian mixture model (GMM) and Gaussian mixture regression (GMR) to regenerate motions. Ng and Russell (Algorithms for inverse reinforcement learning, 2000 [5]) proposed the inverse reinforcement learning (IRL) method to build up an unknown reward function based on the observed trajectories to characterize solutions.