This project is attempting to elucidate the basic principles governing environmental field model synthesis based on the integration of adaptive robot sampling with human decision-making. In so doing, we are addressing two fundamental technical challenges: 1) Model-Based Asset Allocation and 2) Sampling-Based Model Refinement. The overall goal is to couple human decision-making with probabilistic modeling and learning in a decision support system enabling environmental field model discovery and refinement. The first challenge, Model-Based Asset Allocation, involves synthesis of large-scale, low-resolution data with human scientific expertise to make timely, model-informed asset allocation decisions. The second challenge, Sampling-Based Model Refinement, involves small-scale, high-resolution autonomous cooperative selection and execution of robot sampling trajectories. Both challenges involve the handling of multivariate, multi-resolution, temporally evolving fields. The second challenge’s high-resolution local field characterization cyclically feeds and refines the first challenge’s high-level model. We will demonstrate these general principles in the specific domain of coastal ocean exploration, but expect them to be broadly applicable across robotics and computational intelligence.

This material is based upon work supported by the National Science Foundation under Grant Number IIS1124941.