UMR: Unified Motion Retargeting via Learned Point Cloud Correspondence
Loading video
Loading videoUMR learns dense point cloud correspondences as a unified interface for transferring human motion across heterogeneous humanoid embodiments, replacing hand-crafted keypoint mappings with constrained point cloud matching that aligns surface poses and transfers interaction contacts.
Category: humanoid
Author: @Runyi_Ingrid_Yu
Date: 2026-09-11T00:00:00
Duration: 123.297s
Reference: https://hanyang9.github.io/UMR/
