Extended reality lab for Mars experiments (Mars Xlab)

The main objective of this activity was to design and develop an innovative virtual lab for Mars experiments, where AI methods are used to improve orbital imagery resolution, as well to obtain terrain features based on local images from rovers. Two different methods were used to recreate the virtual environment. The first one was dedicated to process low resolution images from orbital information. The main contribution in this sense was the use of AI to increase the resolution of these images. The generated textures were integrated along with Digital Terrain Maps (DTM) to recreate the surface in 3D. The second approach was related to the 3D recreation of high resolution rocks. Rocks were recreated based on a two stages approach, in which AI was used to detect rocks on rover local images to later recreate them in 3D using GAN Neural Networks. Finally, high-resolution textures were obtained from available sources on Internet, placing them on the virtual scenario depending on the NOAH-H deep-learning based terrain classification system [4], first obtaining the segmentation of a partial area of Mars and, second, including the corresponding high resolution textures.