Robotic Navigation based in Statics Patterns using the CMUcam3 Embedded System
AbstractThis paper presents a method based on static patterns for robotic navigation using computer vision techniques implemented on CMUcam3 embedded system, also presents an analysis of computational complexity for this embedded system. The pattern to be determined is a black line with intersections, driving the robotic agent over the line. A segmentation algorithm based in threshold is used after the images acquisition by the CMUcam3, then defines the thinning patterns and then the linear Hough transform is applied to determine the lines, the angles and the type of the intersections. Due to the linear Hough transform is a method that requires high processing, in this project, the range of the angles are limited and the accumulation space is normalized. The results showthat the method developed for the navigation is accurate and reliable, because in 87% of the decision segments it could determine correctly the type of intersection and the correction angle. The CMUcam3 embedded system succeeded to process an image between 0.15 and 0.28 seconds, depending of the type of the intersection.
The texts published in this magazine, as of June of the year 2018, are under a Creative Commons License "Recognition-Non-Commercial-Share Equal" that allows others: