QDGset
收藏资源简介:
QDGset是由巴黎索邦大学和阿利坎特大学的研究团队创建的一个大规模抓取数据集,包含约6200万次6自由度抓取动作和约4万种模拟对象。数据集通过质量多样性(QD)算法生成,显著提高了抓取数据集的生成效率。数据集的创建过程结合了对象网格的变换和迁移学习方法,旨在解决机器人抓取任务中的数据需求问题,特别是为模拟到现实的抓取任务提供高质量的数据支持。
QDGset is a large-scale grasping dataset created by research teams from Sorbonne University Paris and the University of Alicante. It contains approximately 62 million 6-degree-of-freedom grasping motions and about 40,000 simulated objects. The dataset is generated via Quality Diversity (QD) algorithms, which significantly improves the generation efficiency of grasping datasets. Its creation process combines object mesh transformation and transfer learning methods, aiming to address the data demand issues in robotic grasping tasks, particularly providing high-quality data support for sim-to-real grasping tasks.

- 1QDGset: A Large Scale Grasping Dataset Generated with Quality-Diversity巴黎索邦大学,法国国家科学研究中心,智能与机器人系统研究所 · 2024年



