Zusammenfassung
SummaryIndoor environments can typically be divided into places with different functionalities like corridors, rooms or doorways. The ability to learn such semantic categories from sensor data enables a mobile robot to extend the representation of the environment, and to improve its capabilites. As an example, natural language terms like corridor or room can be used to communicate the position of the robot in a more intuitive way. Other tasks, like exploration or localization, can also be carried out by the robot in a better way when semantic information is taken into account.In this thesis, we present a method that enables a mobile robot to classify the different places of indoor environments into semantic classes, and then use this information to extend its representations of the environments. The main idea is to classify the position of the robot based on the current observations taken by the robot. In this work, we use as main observations the scans obtained from a laser range sensor. Each scan is represented by a set of features that encode the geometrical properties of the current position. These features are then used to classify the scan into the corresponding semantic class. The output of the classification is represented by a probability distribution over the set of possible semantic classes. This probabilistic representation permits us to apply further probabilistic techniques to improve the final classification, reducing the number of errors. We also present an extension which enables the robot to include other types of observations in the classification, like camera images.This work additionally introduces several applications of the previous approach in different robotic tasks. First, we will show how the semantic information can be used to extract topological maps from indoor environments. In a second application, we present a method that incorporates transitions between different places when classifying a trajectory taken by a mobile robot. It will also be shown that the semantic information can reduce the time needed by the robot in exploration and localization tasks. Finally, we present the semantic classification of places as part of an integrated robotic system designed for interacting with humans using natural language.