Event
Authors propose STR and STRL methods for humanoid robot motion retargeting
1 reports · 1 sources
Xiongbo Shi and other authors propose human motion retargeting methods STR and STRL for humanoid robot mobile manipulation. STR uses a vision-language model to select relevant motion signals and body parts, then identifies keyframes based on changes in the selected signals, allocating additional solving iterations to keyframes and assigning higher interaction weights to the selected parts, in order to reduce retargeting computational overhead. The authors report that simulation experiments on the Unitree G1 and real-world experiments on the Unitree R1 show that STR improves geometric contact coverage and continuity at lower computational cost.
Report timeline
- STR: A Spatiotemporal Preference Motion Retargeting Method for Humanoid Loco-Manipulation
The authors propose a spatiotemporal retargeting method, STR, which uses a vision-language model to select relevant motion signals and body parts and identifies keyframes accordingly, allocating additional solver iterations to keyframes and higher interaction weights to the selected parts to reduce retargeting computational overhead.
arXiv cs.RO ·