Abstract: | The thesis presents a visual position measurement system that allows an Automated Guided Vehicle (AGV) to recover its absolute position anywhere on a pseudo-random encoded guide-path. It is based on the "window property" of pseudo-random binary sequences. The guide-path consists of an (2$\sp{\rm n}$-1) bit long encoding pseudo-random binary sequence marked on the floor as a sequence of geometric binary symbols ('E' for the binary value 1, and 'H' for the binary value 0). A long portion of the encoded guide-path has to be visible to the camera mounted on the AGV, in order to provide enough information to recover the n-bit pseudo-random window attached to any point in the field of view. The binary contents of this window uniquely identifies the absolute position of the encoded point. Image analysis and pattern recognition algorithms have been developed for the visual recovery of the pseudo-random window. The image processing essentially consists of the following steps: thresholding of the original image, boundary tracing each object in the image, polygon approximation of each boundary, extracting vertices and corners from each polygonal contour, and finally binary-symbol recognition by tree search. An original implementation of the split and merge polygon approximation results in an improvement of the run-time performance of this algorithm. An experimental vision system was implemented to test its performance for AGV absolute position recovery applications. |