Retargeted_AMASS_for_robotics
收藏资源简介:
该项目旨在将AMASS(Archive of Motion Capture as Surface Shapes)数据集中的运动数据重新定位到各种机器人模型上,并开源这些重新定位的数据,以促进机器人学和人机交互的研究和应用。AMASS是一个高质量的人类运动捕捉数据集,SMPL-X模型是生成逼真人类运动数据的强大工具。通过将AMASS中的运动数据适配到不同的机器人模型,该项目希望为机器人训练和人机交互提供更多样化和易于访问的运动数据集。数据集内容包括重新定位的运动文件、使用示例和许可证文件。重新定位的运动文件包括针对Unitree G1机器人的运动数据,这些数据基于Unitree提供的官方开源模型生成,并符合XML文件中定义的关节位置约束。
This project aims to retarget motion data from the AMASS (Archive of Motion Capture as Surface Shapes) dataset onto various robot models, and open-source these retargeted motion data to facilitate research and applications in robotics and human-robot interaction. AMASS is a high-quality human motion capture dataset, while the SMPL-X model serves as a powerful tool for generating realistic human motion data. By adapting the motion data in AMASS to different robot models, this project seeks to provide more diverse and readily accessible motion datasets for robot training and human-robot interaction. The dataset includes retargeted motion files, usage examples, and license files. Specifically, the retargeted motion files cover motion data for the Unitree G1 robot, which are generated based on the official open-source model provided by Unitree and conform to the joint position constraints defined in the XML file.




