Research On Target Detection And Path Identification Method Based On Machine Vision And DSP

Multi-robot system is at the forefront in the field of robotics research and the applicationto agriculture will be innovation. In the relevant research on farmland multi-robot system,work assignment and autonomous work is the key of realizing robots to cooperate. Machinevision is the basic requirements of robots’environmental perception, so the related imageprocessing algorithms is the top priority of the research. Path detection and targetidentification based on monocular vision has the advantage of simple arithmetic and fastdetection, using DSP technology for the hardware foundation has the advantage of goodreal-time, fast processing speed, small size, low power consumption, and so on, so theresearch of this paper has important theoretical significance and practical value.The paper focuses on the machine vision of the multi-robot system, developed a pathdetection method applied to the leading robot based on machin vision and DSP technology,came up with a rapid target identification methoh applied to following robot, including thefollowing work:（1）Setted up the hardware experiment platform, it’s core is the SEED-VPM642whichis developed by Beijing Hezhongda company, the hardware platform used a dedicatedhigh-performance DSP as the processing core, using the idea of modular while the researchprocess, including DSP system（CSL library functions, EMIFA interface, the EDMA, videocodec chip, etc.）initialization, the input and output of video, the DSP core image processingalgorithms, etc., completed the simulation and testing of the entire DSP system.（2）Designed the visual system and the algorithm of target identification for followingrobots，made a red cylinder as the target for detection to simulate the leading robot，extractedthe target by defining color domain from background, after image binarization, identificatedthe target with the neighborhood method, calculated the width a and coordinate value x, usedthe curve fitting tool of Matlab to determine the equation of a and x to solve the distance andthe angle of the target, inserted the equation to DSP to take experiments, experiments turnedout that this algorithm works well.（3）Selected the VPM642development board as the hardware platform of the robot visual system, completed designing the algorithm of path detection（including the forkdetection and the edge detection） for the leading robot, after several experiments, it turns outthat the algorithm has the cherecter of good real-time, constructed the basic frameworkfor visual system of the robots.