TATERS (Tool for Autonomous Terrain Exploration of Remote Spaces)

摘要

This paper reports on the development of TATERS (Tool for Autonomous Terrain Exploration of Remote Spaces), a software framework for planning and risk analysis of single and multirobot traversals. TATERS is designed to aid in mission planning workflows by merging planning methodologies with execution of autonomy on board robotic systems. This is done by developing coarse and high-fidelity simulations as well as tying in features of autonomy stacks into Monte Carlo simulations during pre-mission studies. Further, we investigate and report on methods for curating live data, incorporating into the planning activities, and representing and updating optimal paths on autonomous systems without the need for updating large amounts of data. This is achieved by a data reduction scheme called skeletonization which compresses spatio-temporal cost cube data derived from geographic information system (GIS) data layer sets. The cost cube derivation strategy is designed using cost metrics that can be adjusted and guided by intuition. We identify and focus on three key metrics: scientific merit, traversability, and survivability.