Vision Based Navigation Datasets
收藏资源简介:
Vision Based Navigation Datasets是由欧洲空间局主导,空客防务与空间公司参与创建的数据集,旨在支持基于视觉的导航技术在航天领域的应用。该数据集包含多个子集,涵盖了从月球着陆到人造卫星对接等多种场景,总计超过13万条数据。数据集的创建过程结合了真实图像、实验室模拟和合成图像,确保了数据的多源性和高质量。这些数据集主要用于训练机器学习算法,特别是在姿态估计和光学流算法方面,以解决航天器导航中的精确控制问题。
Led by the European Space Agency (ESA) and co-developed with Airbus Defence and Space, the Vision Based Navigation Datasets are a collection of datasets created to support the application of vision-based navigation technologies in the aerospace field. This collection encompasses multiple subsets covering a wide array of scenarios, from lunar landings to artificial satellite docking, with a total of over 130,000 data entries. The development of these datasets integrates real-world images, laboratory simulations, and synthetic imagery to ensure their multi-source nature and high quality. They are primarily utilized for training machine learning algorithms, especially those focused on pose estimation and optical flow, to resolve precision control challenges in spacecraft navigation.

- 1Training Datasets Generation for Machine Learning: Application to Vision Based Navigation欧洲空间局 · 2024年



