Estimation of Virtual Moon Surface using LiDAR Measurements

In this work, we present the MLS (Moon Lidar Simulation) system—a terrain height map and normal map prediction system for virtual lunar landscapes that uses LiDAR point cloud data as input. When operating the MLS system, we first simulate a virtual LiDAR sensor to sample height information from a 3D terrain and preprocess the data to generate an initial height map. We then leverage a Masked Autoencoder (MAE) structure to predict a normal map based on the height map input. Our system demonstrates its capability to predict occluded terrain areas, and we visualize the resulting height map and normal map in the Unity game engine.