Learning Pose-Aware Action Primitives via Functional Object Canonicalization for Generalizable Robotic Manipulation

Hongli Xu2,3*, Lei Zhang1,3*†, Xiaoyue Hu2*, Boyang zhong2, Kaixin Bai1,3,

Zoltán-Csaba Márton3, Zhenshan Bing2, Zhaopeng Chen3, Alois Christian Knoll2, Jianwei Zhang1

1 Universität Hamburg, 2 Technical University of Munich, 3 Agile Robots SE
* The first three authors contribute equally to this paper
†Corresponding author. lei.zhang-1@studium.uni-hamburg.de

📕Arxiv 📄Supplementary Materials

FunCanon Structure

Functional Alignment, Automatic Trajectory Transfor, FuncDiffuser, and Sim2Real Transfer

Real World Robotic Experiments

In case this work has provided some assistance in your research endeavors, your citation would be greatly appreciated.

@article{TODO}

Page updated

Google Sites

Report abuse

FUNCanon

Learning Pose-Aware Action Primitives via Functional Object Canonicalization for Generalizable Robotic Manipulation

Real World Robotic Experiments