HiFun: A Hierarchical Framework for Efficient Functional Dexterous Manipulation Learning
HiFun: a Hierarchical Framework for Efficient Functional Dexterous Manipulation Learning
This paper presents HiFun, a hierarchical real-world RL framework for functional dexterous manipulation tasks. It decouples control complexity into a functional hand skill for precise contact execution and an arm-hand coordination policy for arm control and hand-skill activation. HiFun achieves 98.3% average success in six real-world tasks within one hour of online training and remains robust to dynamic disturbances.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
training time
1 hours
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
th Dynamic Movement Primitive (DMP) augmentation. Across six real-world tasks on two dexterous hands, HiFun achieves 98.3% average success within one hour of online training and remains robust to dynamic disturbances. The project page is available at https://hly-123.github.io/HiFun/.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org