L1-norm global localization based on a Differential Evolution Filter

Abstract

Global localization methods deal with the estimation of a mobile robot's pose assuming no prior state information about it and a complete a priori knowledge of the environment where the mobile robot is going to be localized. Most existent algorithms are based on the minimization of a L2-norm loss function. However, the use of a L1-norm offers some alternative advantages. The present work explores the use of a L1-norm together with an Evolutive Localization Filter to determine its efficiency when applied to the global localization problem. The algorithm has been tested subject to different noise levels to demonstrate the accuracy, effectiveness, robustness and computational efficiency of the L1-norm approach.