Robot Manipulation Task Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://philippente.github.io/pub/GMR.html
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了从六轴机械臂（Kinova Jaco 2）在模拟和现实操作任务中收集的轨迹样本，其中包括插孔任务。该数据集用于评估生成性运动反射策略相较于先前状态-动作策略的性能和鲁棒性。每个条件共有50个轨迹样本，在多次迭代中进行评估，其任务涉及机器人操作与控制策略学习。
This dataset contains trajectory samples collected from a 6-axis robotic arm (Kinova Jaco 2) during both simulated and real-world manipulation tasks, including the peg-in-hole task. This dataset is used to evaluate the performance and robustness of generative motor reflex strategies compared to prior state-action policies. Each condition includes 50 trajectory samples, and evaluations are carried out across multiple iterations, with the tasks involving robotic manipulation and control strategy learning.
提供机构：
Kinova Robotics
