A doctoral thesis studied how to improve the visual navigation of mobile robots
April 7, 2010
Researcher Ivan Villaverde, Member of the Group of intelligence computational of the Faculty of Informatics of the BASQUE country, has submitted a thesis on the application of algorithms used in the line of research of hyperspectral image analysis. ' On computational intelligence tools for vision based navigation of mobile robots' (techniques of computational intelligence for visual navigation of mobile robots) is the title of this work to study how to improve the special location as well as the resources of small mobile robots for use in the workplace, using techniques that had not proven in the field of Robotics with previously. To this end, it has inspired in an algebraic system, networked computing, which is used in the research of the hiperespectrometría. The system is based on sets of data, instead of numbers, with a particular internal order.
In his research, Villaverde has used two types of basic sensors that would act as the eyes of these mobile robots: optical cameras and 3D cameras range detection. With the intention that these eyes work properly, the researcher has taken into account three basic issues: the location of the robot and its ability to detect their own movements as well as the ability to construct a map of any environment that is unknown. Three questions to which applied techniques as a lattice computation for the autolocalización of the robot in qualitative maps and the detection of Visual beacons with optical cameras. On the other hand, Villaverde has also used a hybrid system that improves the metric location with 3D cameras: evolving techniques and competitive neural networks. The first algorithms are genetic, and the latter, codes that act as the nervous system of humans. Both resemble the mutations and human developments. The researcher has applied these techniques to 3D cameras; mainly to the estimate of changes between 3D views, and at the same time, the presumable movement of the robot.