Statistical Models and Decision Making for Robotic Scientific Information Gathering

Genevieve Flasphohler, S.M., 2018
Yogesh Girdhar, Advisor

Mobile robots and autonomous sensors have seen increasing use in scientific applications, from planetary rovers surveying for signs of life on Mars, to environmental buoys measuring and logging oceanographic conditions in coastal regions. This thesis makes contributions in both planning algorithms and probabilistic model design for autonomous scientific information gathering. Developing robust autonomous systems that enhance our ability to perform exploratory science requires insight and techniques from classical areas of robotics, such as motion and path planning, mapping, and localization, and from other domains such as machine learning, spatial statistics, optimal experimental design and optimization. This thesis demonstrates how theory and practice from these diverse disciplines can be unified to develop online algorithms for autonomous information gathering that are robust to modeling errors, account for spatiotemporal structure in scientific data, and have probabilistic performance guarantees. The resulting algorithms are applied to real marine science datasets collected by the SeaBED autonomous underwater vehicle at the Hannibal Sea Mount in Panama and at the Martha's Vineyard Coastal Observatory off the coast of Cape Cod, USA to perform state-of-the-art unsupervised model learning and autonomous informative sampling.