An embedded vision system for a power transmission line inspection robot

Abstract

Surrounding environment information is significant for an inspection robot to plan its behavior to stride over obstacles. This paper describes a compact embedded monocular vision system for a power transmission line inspection robot. This system involving a DSP processor has superior hardware design and excellent performance on image processing. Emphasis of the image processing software design is the algorithm of image segmentation and straight line extraction. Base on the structure of the transmission lines, an algorithm for obstacle distance estimation is proposed. Experimental results show that good capability on recognition and location can be achieved with this proposed vision system. © 2009 Springer-Verlag Berlin Heidelberg.