Accurate and complete terrain maps enhance the awareness of autonomous robots and enable safe and optimal path planning. Rocks and topography often create occlusions and lead to missing elevation information in the Digital Elevation Map (DEM). Currently, these occluded areas are either fully avoided during motion planning or the missing values in the elevation map are filled-in using traditional interpolation, diffusion or patch-matching techniques. These methods cannot leverage the high-level terrain characteristics and the geometric constraints of line of sight we humans use intuitively to predict occluded areas. We introduce a self-supervised learning approach capable of training on real-world data without a need for ground-truth information to reconstruct the occluded areas in the DEMs. We accomplish this by adding artificial occlusion to the incomplete elevation maps constructed on a real robot by performing ray casting. We first evaluate a supervised learning approach on synthetic data for which we have the full ground-truth available and subsequently move to several real-world datasets. These real-world datasets were recorded during exploration of both structured and unstructured terrain with a legged robot, and additionally in a planetary scenario on Lunar analogue terrain. We state a significant improvement compared to the baseline methods both on synthetic terrain and for the real-world datasets. Our neural network is able to run in real-time on both CPU and GPU with suitable sampling rates for autonomous ground robots. We motivate the applicability of reconstructing occlusion in elevation maps with preliminary motion planning experiments.

Dataset acquisitions devised specifically for robotic planetary exploration are key for the advancement, evaluation, and validation of novel perception, localization, and navigation methods in representative environments. Originating in the Bardenas semi-desert in July 2023, the data presented in this Data Descriptor is primarily aimed at Martian exploration and contains relevant rover sensor data from approximately 1.7km of traverses, a high-resolution 3D map of the test area, laser-induced breakdown spectroscopy recordings of rock samples along the rover path, as well as local weather data. In addition to optical cameras and inertial sensors, the rover features a thermal camera and six force-torque sensors. This setup enables, for example, the study of future localization, mapping, and navigation techniques in unstructured terrains for improved Guidance, Navigation, and Control (GNC). The main features of this dataset are the combination of scientific and engineering instrument data, as well as the inclusion of the thermal camera and force-torque sensors in particular.

The use of autonomous rovers for planetary exploration is crucial to traverse long distances and perform new discoveries on other planets. One of the most important issues is related to the interaction between the rover wheel and terrain, which would help to save energy and even avoid getting entrapped. The use of reconfigurable rovers with different locomotion modes has demonstrated improvement of traction and energy consumption. Therefore, the objective of this paper is to determine the best locomotion mode during the rover traverse, based on simple parameters, which would be obtained from propioceptive sensors. For this purpose, interaction of different terrains have been modelled and analysed with the ExoTeR, a scale prototype rover of the European ExoMars 2020 mission. This rover is able to perform, among others, the wheel walking locomotion mode, which has been demonstrated to improve traction in different situations. Currently, it is difficult to decide the instant time the rover has to switch from this locomotion mode to another. This paper also proposes a novel method to estimate the slip ratio, useful for deciding the best locomotion mode. Finally, results are obtained from an immersive simulation environment. It shows how each locomotion mode is suitable for different terrains and slopes and the proposed method is able to estimate the slip ratio.