Risk Assessment of the Traversability of Subterranean Planetary Environments

Abstract

In the realm of this thesis's work, a risk assessment of the traversability of subterranean planetary environments with challenging terrain was investigated. The Scout Rover, which was developed by the DLR (German Space Agency), has a unique and robust design made for traversing in challenging environments, and through the collaboration with the DLR the Scout Rover could be used as the robotic platform to evaluate the research objective of this thesis's work. In order to assess the risk of these environments it was necessary to first to include the information about height, as many elevated areas are present. Therefore, the map of the terrain is build as a 2.5D Elevation Map [7], which than is transformed into a Traversability Map [26], offering a more broader representation of the terrain. The Traversability Map contains information on the slope and roughness of the terrain with respect to the capabilities of the Scout Rover. In Chapter 5.1 their critical values for slope t_{critical,s} and roughness t_{critical,r}, describing the maximal values for slope and roughness of the terrain the Scout Rover's can handle, are identified as 0.698 and 0.058. The Traversability Map could either be build locally with the Scout Rover's Point Cloud pipeline without any prior information about the terrain, or the height-map of the terrain could be used directly to create a global Traversability Map. The accuracy of the estimation of t_s and t_r of the local Traversability Map, in Chapter 5.2 is validated against the global Traversability Map. Therefore, for three different regions in both Traversability Maps, the values for t_s and t_r were extracted, and their error was calculated using the two metrics RMSE (Equation 5.2) and MAE (Equation 5.3). Thereby, the applied contact dynamics of the Scout Rover's spokes lead to a more realistic locomotion of the Scout Rover, hence a more representative local Traversability Map was build. In the local Traversability Map the estimation of the value for the traversability analysis of the Scout Rover with respect to the slope of the terrain t_s, relatively small values were estimated for RSME and MAE, so that a high accuracy for the local Traversability Map for the slope could be validate. However, the building of the local Traversability Map with the Scout Rover for a rough terrain not for every grid-cell a valid number was estimated. Hence, two classification for the invalid grid-cells in the local Traversability Map were evaluated to validate the accuracy of the local Traversability Map with respect to t_r, which are consisting of either assuming the worst roughness for all invalid grid cell, which is the critical roughness for the Scout Rover, or considering only valid grid-cells. Finally, to assess the risk of traversing the terrain, path-planning was performed using Art Planner [25] for legged robots after it was adjusted for the Scout Rover, described in Chapter 3.3.1. Also, the information about the traversability analysis of the Scout Rover was incorporated into the path planning objective by designing a new cost function c_{total} described in Chapter 5.3.1 and evaluating it's design first on a simple terrain in Chapter 5.3.2. Lastly, the risk assessment of the traversability of subterranean planetary environments were evaluated on the DLR's Moon and Mars Testsuite Traversability, discussed in Chapter 5.3.3. Thereby, it could be demonstrated for a path (Path-1) in the Traversability Map of the DLR's Moon and Mars Testsuite, that the Scout Rover's cost or effort of traversing through a rough terrain to reach an inclined area decreases by 40% if the safer course of the path was chosen, avoiding the rougher and inclined areas as much as possible estimated through the traversability analysis of the Scout Rover.