Abstract
The main objective of this paper is to address the mobile robot localizationproblem with Triplet Convolutional Neural Networks and test their robustnessagainst changes of the lighting conditions. We have used omnidirectional imagesfrom real indoor environments captured in dynamic conditions that have beenconverted to panoramic format. Two approaches are proposed to addresslocalization by means of triplet neural networks. First, hierarchicallocalization, which consists in estimating the robot position in two stages: acoarse localization, which involves a room retrieval task, and a finelocalization is addressed by means of image retrieval in the previouslyselected room. Second, global localization, which consists in estimating theposition of the robot inside the entire map in a unique step. Besides, anexhaustive study of the loss function influence on the network learning processhas been made. The experimental section proves that triplet neural networks arean efficient and robust tool to address the localization of mobile robots inindoor environments, considering real operation conditions.