Architecture of a surface exploration traverse analysis and navigational tool

Human path planning has typically been performed manually, especially within the context of planetary Extravehicular Activities (EVAs) — whether historically (Apollo program) or simulated (e.g., NASAs Desert Research and Technology Studies (DRATS)). However, human ability to explore and extract relevant information for path planning from datasets such as birds-eye imagery or hillshaded terrain is limited. This can lead to plans that are more time and energy intensive than anticipated, or plans that are revealed to be hard to traverse during mission execution. When combined with time delays on Mars, where the expert path planners would be on Earth, there is a clear need for rapidly re-planning under contingencies. Surface Exploration Traverse Analysis and Navigational Tool (SEXTANT) addresses these challenges through automation of the traverse planning process. The tool has been used on several case studies and most recently it was successfully adopted and deployed during the field-based Mars Exploration Analog mission: NASA BASALT (Biologic Analog Sciences Associated with Lava Terrains) research project. This paper describes SEXTANTs software architecture, and critical changes that have been made for it to adapt to a real use. It brings together three components: an environmental model (terrain data), an explorer model (with energetic cost functions) and a solver or path planner. Notably one solver for cost functions in general, based on the A∗ algorithm is discussed, and the paper goes into more detail on how to use the A∗ algorithm with large map datasets, planning when raw data resolution varies, and planning when computational performance is paramount. The latter point is especially critical for real-time re-planning in the field under potential contingencies. Finally the paper brings forth how to use SEXTANT in practicality, and the flexibility it offers to interface with most other applications.