3D Mapping of an Unknown Environment by Cooperatively Mobile Robots

摘要

Some of the important future applications ofmobile robot systems are autonomous indoor and outdoormeasurements of building, factories and objects in threedimensional view. A goal of such measurement is to provide uswith a detailed map of the environment with the interestingcharacteristics. This map can then be transferred into a model,which represents the measuring objects. By following anautomated proceeding, an autonomous measurement robotcould be useful in order to extract independently the map andthe model of the environment. The environmental model canthen be directly used by the autonomous mobile robots fornavigation. A further increase of the effectiveness in mapproduction can be achieved, if the environment is mapped byseveral robots, or whole robot fleets. Such an approach can be for example, by reducing the exploration time. This paper develops the framework for 3D mapping using multimobilerobots. Two mobile robots equipped with differentsensors capabilities are used. The main contribution of thepaper is to autonomously build a 3D map of indoorenvironments within a good exploration time by using multimobilerobots. Employing multiple autonomous robots withdifferent types of sensors, two different algorithms arepresented in this paper. The first is an algorithm for naturalfeature extraction using stereo camera in order to build a 3Dfeature-based map. The second is an algorithm to extractgeometrical features from range images in order to build a 3Dmodel of the environment. The algorithms and the frameworkare demonstrated on an experimental testbed that involves ateam of two mobile robots. One of them is working as a masterequipped with stereo camera. Whereas the second is involvedas a slave equipped with a rotated laser scanner sensor.