The greatest single source of risks for Mars rovers is terrain. These risks are currently managed by a labor-intensive process in which rover operators carefully examine the terrain and plan a path to avoid any potential hazards. This poses a challenge, particularly for the operation of an MSL (Mars Science Laboratory)-class rover, because it must be very risk-averse in order not to lose the asset, while it already requires a significant amount of labor due to the complexity of the rover. Hence, it is important to develop a software tool that helps operators to detect and avoid terrain hazards efficiently and reliably.

A ground-based Risk-aware Mars Rover Operation Tool mitigates risks from terrain without relying on increased labor by automatically identifying hazards on the terrain, evaluating their risks, and suggesting operator safe path options that avoid potential risks while achieving specified goals.

The tool will be built upon two existing technologies. The first is a machine learning-based terrain classification capable of identifying potential hazards, such as pointy rocks and soft terrains, from images. It is based on the random forest algorithm, which has been successfully demonstrated. The classifier was trained by 65+ training data labeled by human experts. The other technology is risk-aware path planning, which suggests safe paths in consideration of terrain types, slopes, and positive and negative obstacles. It is built upon the rapidly exploring random graph (RRG) and A* search algorithms.

The machine learning approach enables capture of the art of human experts, as well as to incrementally improve performance as a mission accumulates data. The multi-objective optimization capability of the path planner enables users to consider various risk factors (terrain type, tilt, wheel placement, position uncertainty, etc.) and generate multiple suggestions. An interactive GUI interface allows users to work collaboratively with the algorithm. Data export capability to rover operation tools [MarsViewer/RSVP (Robot Sequencing and Visualization Program)] makes the algorithm immediately applicable to MSL/MER (Mars Exploration Rover) operations. Finally, this is a software-only solution, meaning that the algorithm uses only the data currently available to MSL/MER [camera images and DEM (digital elevation model)]; no additional hardware is required.

This technology is immediately applicable to MSL and MER operations as well as future planetary rovers. It can reduce the risk of the loss of rovers by preventing oversight of terrain-related risks, as well as reducing operation cost by partially unloading the effort of Rover Planners (RPs) and Surface Properties Specialists (SPSs) to detect and avoid hazards. The technology is also applicable to onboard hazard detection and avoidance, enhancing Autonav to be fully aware of terrain hazards. It can extend driving distance per Sol by enabling rovers to safely go beyond the sight.

This work was done by Masahiro Ono, Thomas Fuchs, Amanda C. Steffy, Nicholas T. Toole, Mark W. Maimone, and Jeng Yen of Caltech for NASA’s Jet Propulsion Laboratory. This software is available for commercial licensing. Please contact Dan Broderick at This email address is being protected from spambots. You need JavaScript enabled to view it.. Refer to NPO-49679.