ABSTRACTS OF ARTICLES OF THE JOURNAL "INFORMATION TECHNOLOGIES".
No. 8. Vol. 26. 2020

DOI: 10.17587/it.26.472-481

V. A. Bimakov, Director, e-mail: robint@mail.ru, E. V. Bimakov, Department Head, e-mail: robint@mail.ru, Ltd "Voxel technique", Izhevsk, Russian Federation

Improving the Speed and Quality of Analysis of 3D-Scenes Scanned by Sensors of an Autonomous Robot

The issues of construction and functioning of computer vision systems (CVS) intended for use in control systems of super-adaptive autonomous robots (ARCS) are considered. The concept of the quality of the geometric analysis of the scene performed in the CVS of AR is defined. A generalized algorithm for the functioning of the CVS of AR is presented, based on the use of ultra-high-speed hardware-software technology for scene processing (the so-called voxel HSTPS). On the example of the implementation of individual procedures of this algorithm, the advantages of using voxel HSTPS to build super-adaptive ARs based on it based on a geometrically complex and geometrically rapidly changing external environment are demonstrated.
Keywords: computer vision, adaptability of autonomous robots, computing technologies for processing ÇD-scenes, geometric primitives, geometric models of ÇD-objects, recognition of ÇD-objects

P. 472–481

To the contents