Algorithm of data fusion for visual data and range meter measurements for the problem of simultaneous localization and mapping (SLAM)

М.П. Мухіна

Abstract

 The analysis of existing methods for detection of characteristic objects in environment is performed for the problem of Simultaneous Localization And Mapping (SLAM). The variant of integrated use of threshold methods and edge detection methods is considered for building the contours of contrast object. The algorithm of integrated use of range measurements is developed to reject the contours, which are not coincided with real physical objects, that is, those which reflect signal. It is shown that the developed algorithm sufficiently increases the accuracy and stability of detection of environment features, suitable for use in SLAM.

Keywords

позиціонування; картографування; пороговий метод; метод виділення країв; супервізорні прибори; комплексування даних

References

Berkeley Segmentation Dataset – база зображень університета Берклі, [Електронний ресурс]. – режим доступу: http://www.eecs.berkeley.edu/Research/Projects/CS/vision/ grouping/segbench

Durrant-Whyte, H Simultaneous localization and mapping: part I / Durrant-Whyte, H. Bailey, T. // Robotics & Automation Magazine, IEEE. – Vol. 13. – 2006. – Р. 99 – 110.

Andrew J. Davison, Yolanda Gonzalez Cid, Nobuyuki Kita Real-time 3D slam with wide-angle vision // 5th IFAC/EURON Symposium on Intelligent Autonomous Vehicles, Instituto Superior Técnico, Lisboa, Portugal, July 5 – 7, 2004. – P. 120 – 124.

Левашкина А. О., Поршнев С. В. Исследование супервизорных критериев оценки качества сегментации изображений // Известия Томского политехн. ун-та. 2008. Т. 313, № 5. – С. 28 – 33.

Rafael C. Gonzalez, Richard E. Woods “Digital Image Processing” Second Edition, ISBN 0-201-18075-8, Copyright ©2002 by Prentice-Hall, Inc. – 716 p.

Full Text: PDF