V. A. Bobkov, Dr. Tech. Sc., Chief Scientific Researcher, e-mail: bobkov@iacp.dvo.ru, V. P. May, Cand. Tech. Sc., Leading Scientific Researcher, e-mail: may@iacp.dvo.ru, The Institute of Automation and Control Processes, Vladivostok, 690041, Russian Federation

Visual Navigation of Autonomous Underwater Robot with Loop Closing

The method of visual navigation of an autonomous underwater robot (AUV) for conditions of local maneuvering is described in article. The method aims to increase the accuracy of robot localization through generation and the using of virtual network for coordinate binding. The algorithms that implement the proposed method are based on the use of visual odometry and the proposed algorithms for re-visited places. The virtual coordinate binding network consists of the binding points generated at the initial stage of the AUV movement. A binding point refers to a piece of data associated with a particular AUV position. This piece of data includes the position coordinates, a stereo pair of images visible by the camera from a given position, and a sequence of local geometrical transformation matrices accumulated by a given moment in time. It is assumed that the binding points have a high accuracy of coordinates since the method of visual navigation does not accumulate a large error with short movements of the AUV. At the stage of using the virtual network for coordinate binding a search and binding to the nearest binding point are performed. Improving the accuracy of AUV navigation is achieved by replacing a long sequence of local transformations associated with the current position with a shorter sequence associated with the binding point. Estimates of the effectiveness of the method on virtual scenes are obtained.
Keywords: visual odometry, autonomous underwater vehicle, navigation, loop closing, virtual coordinate referencing network

P. 602–607

Acknowledgements: This work was supported by the Russian Foundation for Basic Research (project No. 18-07-00165), "Priority scientific research studies for comprehensive development of the Far-Eastern Division of the Russian Academy of Science" Program (project No. 18-5-014) and the Programs of the RAS Presidium "Fundamental problems of tackling complex practical tasks using supercomputers" (project "Supercomputer modeling of dangerous oceanic and other natural phenomena and control of technogenic objects").