| FULL TEXT IN RUSSIAN
Mekhatronika, Avtomatizatsiya, Upravlenie, 2018, vol. 19, no. 8, pp. 552—560
DOI: 10.17587/mau.19.552-560
A Three-Dimensional Version of the Hough Method in the Reconstruction of the External Environment and Navigation
V. P. Noskov, noskov_mstu@mail.ru, I. O. Kiselev, kiselev201@rambler.ru, Bauman Moscow State Technical University, Moscow, 105005, Russian Federation
Corresponding authors: Noskov Vladimir P., Ph. D., Bauman Moscow State Technical University, Special robotics and mechatronics department, NIISM sector head
Accepted on April 27, 2018
The prevailing issues regarding the three-dimensional (3-D) reconstruction of the industrial-urban environment and navigation model can be solved by using the method of selecting linear objects (straight lines and planes) in cloud points. The analysis of the three-dimensional version of the Hough method for selecting planar objects from a cloud of points has been done. A high-speed algorithm which is a development of the Hough method, based on a two-stage transformation of the initial data, taking into account their linear structuring into a three-dimensional parameter space has been proposed. The linear structuring of the input data generated by 3D laser sensors allows efficient selection of planar objects: first to find subsets of points belonging to line segments in the scanning planes, and then to find subsets of line segments belonging to one flat object. Comparative estimates of the computation volumes for the three-dimensional version of the Hough method and the proposed two-stage algorithm are obtained. Estimates give a quadratic and linear dependence on the same parameter, respectively, which provides for real 3D images increased performance in the second case by two orders of magnitude. The effectiveness of the proposed algorithm is confirmed by the results of the operation of the corresponding software and hardware in real environmental conditions. The obtained results of theoretical and experimental studies allow us to conclude that the created algorithm and software-hardware tools provide a transition in real time (in the rate of motion of control objects) from large volumes of initial visual ranging information to semantic information and navigation models. Formed models in an explicit and compact form contain geometric data about the external environment and navigation data about the sensor (control object). The possibility of forming semantic information and navigation models is based on data obtained from on-board sensors and on-board calculators in real time which allows for the present time to solve ongoing tasks for autonomous traffic control of AGV and UAV in the industrial-urban environment and buildings.
Keywords: point cloud, linear primitives, Hough method, far-ranging 3D-image, linear structuring of initial data, geometric and semantic models, navigation
Acknowledgements: This article was prepared with the financial support of Russian Foundation for Basic Research (project no. 16-29-04178 ofi-m).
For citation:
Noskov V. P., Kiselev I. O. A Three-Dimensional Version of the Hough Method in the Reconstruction of the External Environment and Navigation, Mekhatronika, Avtomatizatsiya, Upravlenie, 2018, vol. 19, no. 8, pp. 552—560.
DOI: 10.17587/mau.19.552-560
To the contents |