Basic Research, Based On Binocular Stereo Vision, Automated Guided Vehicle Navigation

Detailed studies are made on the binocular stereo vision-based navigation of AGV in this paper. These studies aim at improving the intelligent level of AGV, in order to meet the need of development of modern manufacturing industry.On the basis of analysis of composing and theory of AGV based on binocular vision navigation, the placement of binocular vision sensors and passive infrared rangefinder sensors were determined. By analyzing of the method and theory of camera calibration, a camera calibration method based rectangular module was proposed in this paper. A following test proved its high degree of accuracy, which reduced amount of work and improved the real-time character of navigation system. According to the demand of binocular stereo vision-based navigation and avoidance, characteristic point-based matching method was determined. The function was implemented under the VC++ 6.0 developing environment, such as image pre-processing, characteristic point subtraction and matching. The method of space lattice 3-D reconstruction is decided on the basis of stereo correspondence. Using OpenGL three dimensional graph software packages, a following simulation experiment carried on, which indicated that the method met the need of AGV navigation. At last grid-based search algorithm was proposed, and BP nerve network-based algorithm of local path planning was used too, which improved the navigation efficiency of AGV. The simulation software of QDU-II AGV has been developed and dynamic simulation of path planning was achieved successfully.