METHODS OF VISUAL NAVIGATION FOR MOBILE ROBOTS AND CARTOGRAPHIC MODELS FOR THE EXTERNAL ENVIRONMENT
J. S. Pershina1, S. Ya. Kazdorf1, A. V. Lopota2
1Novosibirsk State Technical University, prosp. K. Marksa 20, Novosibirsk, 630070 Russia
2Central Research Institute of Robotics and Technical Cybernetics, Tikhoretskii pr. 21, St. Petersburg, 194064 Russia
Keywords: визуальная навигация, свёрточные нейронные сети, семантическая сегментация, когнитивная карта, visual navigation, convolutional neural networks, semantic segmentation, cognitive map
Abstract
Modern methods of visual navigation for mobile robots are considered. A hierarchical representation structure of the surrounding space corresponding to the hierarchical organization of the mobile robot control system is proposed. Modern approaches to building cartographic models are presented. Their development will bring the navigation system closer to that created by the human intellect and combines vision and a semantic view of the world within cognitive maps.
|
