视觉导航AGV定位与路径规划技术研究

发布时间：2018-03-30 09:55

本文选题：视觉导航AGV　切入点：视觉定位　出处：《华南理工大学》2016年硕士论文

【摘要】：工业自动化和智能化水平的提高对物流系统提出了更高的要求。自动引导小车(AGV)作为现代物流系统的重要组成部分,现已得到广泛的应用。在众多AGV导航方式中,视觉导航方式在路径变更、智能避障、功能拓展等方面有较大优势,发展前景广阔。本文针对AGV视觉导航技术开展了研究工作,主要研究内容如下:首先,在分析视觉导航AGV发展现状及视觉导航关键技术的基础上,确定了本文的研究重点为定位技术和路径规划技术。从多个方面对比分析了固定路径视觉导航方式和自由路径视觉导航方式,提出了本文的视觉导航方案并对硬件进行了选型和调试。其次,对视觉定位算法进行了研究。根据视觉定位流程,对比分析了多种特征匹配算法的匹配性能,确定了ORB算法作为本文的特征匹配算法,并利用视觉词典算法提高图像匹配速度。同时采用贝叶斯滤波算法建立了图像匹配概率模型,提高了匹配鲁棒性。结合随机抽样一致性算法和最小二乘法计算匹配图像间的相对位姿。随后通过实验对视觉定位算法进行了多方面评估。再次,基于AGV自由路径导航方式,从全局路径规划和局部路径规划两方面开展了研究。为获取运行时间最短路径,利用结合角度的代价评估方法和基于移动全向性的节点剔除方法,提出了一种改进JPS(Jump Point Search)全局路径规划算法。同时为解决局部极小问题,引入沿墙策略,提出了一种改进动态窗口局部路径规划算法。并通过仿真实验验证了两种算法的可行性。最后,搭建了AGV视觉导航系统,并在工业现场开展了实验,验证了本文提出的视觉定位算法和路径规划算法具有较高的准确性和鲁棒性。从而表明基于本文视觉定位算法和路径规划算法的视觉导航系统可实现AGV的自主导航。
[Abstract]:As an important part of modern logistics system, automatic guided trolley (AGV) has been widely used in many AGV navigation methods. Visual navigation has great advantages in path change, intelligent obstacle avoidance, function expansion and so on. In this paper, the AGV visual navigation technology research work is carried out, the main research contents are as follows: first, Based on the analysis of the development of visual navigation AGV and the key technologies of visual navigation, The research focus of this paper is location technology and path planning technology. The visual navigation mode of fixed path and the visual navigation mode of free path are compared and analyzed from many aspects. In this paper, the visual navigation scheme is proposed, and the hardware is selected and debugged. Secondly, the visual location algorithm is studied. According to the visual positioning flow, the matching performance of various feature matching algorithms is compared and analyzed. The ORB algorithm is used as the feature matching algorithm in this paper, and the visual dictionary algorithm is used to improve the image matching speed. At the same time, the Bayesian filtering algorithm is used to establish the image matching probability model. Combining random sampling consistency algorithm and least square method to calculate the relative orientation of matching images, then the visual location algorithm is evaluated in many aspects through experiments. Thirdly, based on the AGV free path navigation, the matching robustness is improved. The global path planning and local path planning are studied in this paper. In order to obtain the shortest path of running time, the cost evaluation method based on combination angle and the node elimination method based on moving omni-direction are used. An improved JPS(Jump Point search algorithm for global path planning is proposed. In order to solve the problem of local minimization, a wall strategy is introduced. An improved dynamic window local path planning algorithm is proposed, and the feasibility of the two algorithms is verified by simulation experiments. Finally, the AGV visual navigation system is built, and the experiments are carried out in the industrial field. The proposed visual location algorithm and path planning algorithm are proved to be accurate and robust, which indicates that the visual navigation system based on the visual location algorithm and the path planning algorithm can realize the autonomous navigation of AGV.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.41

【相似文献】