Research On The Intelligent AGV Visually-oriented Based On Image Processing

With the widely application of Automated Guided Vehicle (AGV), the scheduling of AGV becomes the linchpin to an efficient logistics system. The intention of this thesis is to design an image processing technology based on FPGA and DSP to achieve fast and accurate vision-guided, which is dedicated of AGV intelligent vision-guided navigation system.With a comprehensive analysis of present vision-guided research, this dissertation pointed out the problems of vision-guided AGV technology in practical application, proposed a collectivity scheme of the special dual-core architecture image capture and processing system based on FPGA and DSP, for vision-guided AGV. Dual-core architecture fully reflects the FPGA's parallel processing capabilities and competitive advantage of DSP, then improved image processing speed.This thesis dedicated to the research of dual-core image processing algorithms based on FPGA and DSP, and lay heavy stress on the FPGA image pre-processing algorithms and dual-core high-speed communication mechanisms. Designed Parallel multi-channel pulse coupled neural network algorithm to achieve fast image binary segmentation, filtering and other processes, and used dual-port memory embeded in FPGA to achieve high-speed image data transferred from EDMA bus, which effectively improved the image processing and recognition control's efficiency.This thesis has proved that the visual system's scheme and the image processing algorithm proposed in the thesis has greatly optimized the performance of AGV vision navigation through the simulation on Matlab, ModelSim and practical experiments add to the theoretical basis, which improved the system speed and the accuracy of identification and positioning, providing a theoretical and technical basis for the AGV application in practical production.Finally, the whole work is summarized and some suggestions are proposed to further study.