arXiv Robotics — research abstracts· Dechen Gao, Yue Yang, Ben Abbatematteo, Nathan Godwin, Pengcheng Wang, Roger Boldu, Steven Man, Zhiyang Dou, Chuan Qin, Sho Nakagome, Eric Whitmire·· 10 days agoEditorial score60
Generative Neural Retargeting for Human-to-Robot Dexterous Manipulation
Generative Neural Retargeting for Human-to-Robot Dexterous Manipulation
Summary
This research proposes Generative Neural Retargeting (GNR), a method that enables efficient and scalable retargeting of human demonstrations to robots for dexter, by sampling from a low-dimensional manifold shared across demonstrations. GNR outperforms traditional methods like MPC, requiring only 8.5% of the samples and achieving a success rate of 56.20%.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
Source：arXiv Robotics — research abstracts · arxiv.org