Terrain Aware Traverse Planning for Mars Rovers

NASA is proposing a Mars Sample Return mission, to be completed within one Martian year, that will require enhanced autonomy to perform its duties faster, safer, and more efficiently. With its main purpose being to retrieve samples possibly tens of kilometers away, it will need to drive beyond line-of-sight to get to its target more quickly than any rovers before. This research proposes a new methodology to support a sample return mission and is divided into three compo-nents: map preparation (map of traversability, i.e., ability of a terrain to sustain the traversal of a vehicle), path planning (pre-planning and replanning), and terrain analysis. The first component aims at creating a better knowledge of terrain traversability to support planning, by predicting rover slip and drive speed along the traverse using orbital data. By overlapping slope, rock abundance and terrain types at the same location, the expected drive velocity is obtained. By combining slope and thermal data, additional information about the experienced slip is derived, indicating whether it will be low (less than 30%) or medium to high (more than 30%). The second component involves planning the traverse for one Martian day (or sol) at a time, based on the map of expected drive speed. This research proposes to plan, offline, several paths traversable in one sol. Once online, the rover chooses the fastest option (the path cost being calculated using the distance divided by the expected velocity). During its drive, the rover monitors the terrain via analysis of its experienced wheel slip and actual speed. This information is then passed along the different pre-planned paths over a given distance (e.g., 25 m) and the map of traversability is locally updated given this new knowledge. When an update occurs, the rover calculates the new time of arrival of the various paths and replans its route if necessary. When tested in a simulation study on maps of the Columbia Hills, Mars, the rover successfully updates the map given new information drawn from a modified map used as ground truth for simulation purposes and replans its traverse when needed. The third component describes a method to assess the soil in-situ in case of dangerous terrain detected during the map update, or if the monitoring is not enough to confirm the traversability predicted by the map. The rover would deploy a shear vane instrument to compute intrinsic terrain parameters, information then propagated ahead of the rover to update the map and replan if necessary. Experiments in a laboratory setting as well as in the field showed promising results, the mounted shear vane giving values close to the expected terrain parameters of the tested soils