Synthetic Lunar Terrain (SLT)
收藏资源简介:
合成月球地形(SLT)数据集由澳大利亚机器学习研究所创建,旨在模拟月球表面的极端光照条件,特别适用于训练和评估神经形态视觉算法。该数据集包括来自事件驱动和传统RGB摄像机的多模态捕捉,以及高分辨率3D激光扫描,用于深度估计。SLT数据集的创建过程涉及在模拟月球环境的测试床上进行数据采集,使用定制的模拟月壤材料,并通过手动操作摄像机进行数据记录。该数据集主要应用于月球探测任务中的导航和着陆系统,以提高其感知能力。
The Synthetic Lunar Topography (SLT) dataset was created by the Australian Institute of Machine Learning to simulate the extreme lighting conditions on the lunar surface, and is specifically designed for training and evaluating neuromorphic vision algorithms. This dataset includes multimodal captures from event-based and conventional RGB cameras, as well as high-resolution 3D laser scans for depth estimation. The development of the SLT dataset involved data collection on a testbed simulating a lunar environment, using custom simulated lunar regolith materials, with data recorded via manual camera operation. It is primarily applied to navigation and landing systems in lunar exploration missions to enhance their perceptual capabilities.




