多维特征下的AGV路径选择与导航研究

发布时间：2018-01-10 09:11

本文关键词：多维特征下的AGV路径选择与导航研究　出处：《南京大学》2017年硕士论文　论文类型：学位论文

更多相关文章： AGV 视觉导航 机器视觉 图像处理

【摘要】：AGV是一种以电力为动力,具有一定负载能力,可以通过车载控制系统实现定位、避障和路径规划的智能设备,通过预设轨迹的方法自动引导设备行进和停止,从而达到自动引导传输的工作目的,极大程度上满足了柔性制造和自动化物流转运的需求。近几年,随着机器视觉技术的成熟,通过视觉来进行导航是AGV的一个研究热点,目前通过特定单一特征可以较好的实现AGV自主导航,如循线导航,但是这种导航方式功能单一,在直线路径导航中能有很好的效果,此外图像内容信息量丰富,探测范围广,机器视觉在物体检测和目标跟踪上有着不可比拟的优势,但是运算量较大,所以如何利用单一视觉传感器又好又快地实现路径的选择和目标定位,是当下AGV导航的研究热点。本文通过图像处理手段,对AGV图像传感器传回的路面图像处理分析,识别出路径线,依据直线特征计算车体与路径线的偏向角度,根据相机成像原理与机器视觉相关知识,建立图像平面与路面的对应关系,依据单应性实现对图像中任意两点的测距。为了解决传统AGV行驶路径单一、工作流程固定的问题,本文提出基于识别文字路标的路径选择方案,在分支路径上设置文字路标,在AGV运动过程中,定位图像中的文本区域并识别,与预置路径节点比对选择路径。
[Abstract]:AGV is a kind of electric power, has a certain load capacity can be achieved by positioning the vehicle control system, intelligent obstacle avoidance and path planning method, through the preset trajectory automatic guide device and stop moving, so as to achieve the automatic guided transmission work to great extent to meet the flexible manufacturing and automation of logistics transport demand. In recent years, with the development of the machine vision technology, through visual navigation is a research hotspot in AGV, can achieve the better AGV autonomous navigation through a specific single feature, such as navigation, but the function of the navigation mode is single, can have a good effect on the straight path navigation, in addition image content rich information, wide range of detection, machine vision has the incomparable superiority in object detection and target tracking, but the large amount of computation, so how to make use of a single The visual sensor and to achieve fast and good selection and target path, is the research focus of the present AGV navigation. Through image processing methods, analysis of pavement image processing AGV image sensor to the identification of outlet diameter, based on linear feature computes the path line and the deflection angle, according to the camera imaging principle of machine vision and related knowledge and the relationship between the image plane and the pavement, according to the single should achieve ranging to any points in image. In order to solve the traditional AGV single path, work flow fixed the problem, this paper put forward the path recognition based on landmark text, set the text on road signs in the branch path, in the AGV process, image positioning in the text area and recognition, path and preset path node than the choice.

【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP23

【相似文献】