用高精度月面模型实现实时月球着陆仿真，助力自主导航系统研发。
High performance Lunar landing simulations
- 基于多分辨率数字高程模型与程序化纹理生成，实现真实月面渲染
- 集成20米分辨率全球月面数据，支持连续实时渲染数TB级数据
- 为未来月球任务的视觉导航算法提供高效仿真平台
自主精准月球着陆依赖视觉传感器。计算机视觉算法通过合成仿真进行设计、训练和测试。多个国家发射的月球轨道器已生成高质量地形模型，全球分辨率可达数十至数百米，局部可达数米。SurRender是一款强大的模拟软件，可充分利用这些数据实现光线追踪。新增功能包括多分辨率数字高程模型（DEM）融合与程序化纹理生成。系统整合了20米分辨率的全球月面模型，包含数TB数据，可连续、实时渲染。该仿真器将成为未来月球任务开发的重要工具。
原文摘要 · Abstract (English)
Autonomous precision navigation to land onto the Moon relies on vision sensors. Computer vision algorithms are designed, trained and tested using synthetic simulations. High quality terrain models have been produced by Moon orbiters developed by several nations, with resolutions ranging from tens or hundreds of meters globally down to few meters locally. The SurRender software is a powerful simulator able to exploit the full potential of these datasets in raytracing. New interfaces include tools to fuse multi-resolution DEMs and procedural texture generation. A global model of the Moon at 20m resolution was integrated representing several terabytes of data which SurRender can render continuously and in real-time. This simulator will be a precious asset for the development of future missions.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。