Intelligent Planning for Sensing and Information Collection (AFOSR)
The goal of this research is to develop intelligent planning theory that provably improves sensing/estimation performance in order to enable long term, robust operations in autonomous vehicles.
1. Hypothesis based Estimator: develop estimation theory which maintains multiple hypotheses of estimator instantiations, thus allowing robustness, memory, accuracy, and the ability to reason.
2. Representations of Perceptual Uncertainties: study different representations of perceptual uncertainties for intelligent planners which trade between accuracy and computational tractability.
3. Optimization based Planning over Perceptual Uncertainty: develop planning theory based on these perceptual uncertainty representations that provably improves estimation performance for information based missions.