Quantifying Uncertainty Towards Information-Centric Unmanned Navigation

Abstract

Abstract — Highly imperfect, inconsistent information and incomplete a priori knowledge introduce uncertainty in sensor-centric unmanned navigation systems. Understanding and quantifying uncertainty yields a measure of useful information that plays a critical role in several robotic navigation tasks such as sensor fusion, mapping, localization, path planning, and control. In this paper, within a probabilistic framework, we demonstrate the utility of estimation- and informationtheoretic concepts towards quantifying uncertainty using entropy and mutual information metrics in various contexts of unmanned navigation via experimental results.