基于双目视觉的机器人路径规划研究

发布时间：2018-03-26 18:03

本文选题：机器人路径规划　切入点：人工势场法　出处：《沈阳理工大学》2017年硕士论文

【摘要】：本文对基于双目视觉的机器人路径规划进行了研究,该课题包含两大块内容,它们分别为基于双目视觉的机器人定位和机器人路径规划方法研究。首先,本文对机器人研究领域中的定位技术和路径规划进行了研究,详细介绍了双目视觉和路径规划的研究现状,接着分析了摄像机成像的针孔模型和畸变模型以及摄像机的标定方法,在此基础上,详细介绍了张氏标定法的原理,并且在双目测试平台上对摄像机进行了标定。其次,对图像特征的提取与匹配进行了比较详细的介绍,包括图像特征及其提取方法、图像匹配及其匹配方法,在此基础上,着重研究了SURF特征提取方法的原理,并对其做了相应的特征提取和匹配实验,针对匹配过程中产生的误匹配问题,本文采用了由RANSAC算法和平均斜率方法组合成的“二步法”策略对存在误匹配的特征点对进行了剔除。再次,对双目视觉的分类、视差测距法、图像特征点跟踪方法和机器人运动参数估计方法进行了详细的介绍,并对双目视觉定位的方法进行了实验验证。然后,对人工势场法的机器人路径规划方法进行了研究,介绍了传统人工势场法的实现及其缺点,针对传统人工势场法存在局部极小点的问题,提出了本文的修改斥力方向和自主建立虚拟目标牵引点相结合的改进算法,并对改进后的算法进行了实验。最后,对本课题已做工作进行了总结,并对后续工作内容作了展望。
[Abstract]:In this paper, the robot path planning based on binocular vision is studied, which includes two parts: robot location based on binocular vision and robot path planning. In this paper, the localization technology and path planning in the field of robot research are studied, and the research status of binocular vision and path planning is introduced in detail. Then, the pinhole model and distortion model of camera imaging and the camera calibration method are analyzed. On this basis, the principle of Zhang's calibration method is introduced in detail, and the camera is calibrated on the binocular test platform. The extraction and matching of image features are introduced in detail, including image features and their extraction methods, image matching and matching methods. On this basis, the principle of SURF feature extraction method is emphatically studied. The corresponding feature extraction and matching experiments are done to solve the mismatch problem in the course of matching. In this paper, the "two-step method" strategy, which is composed of RANSAC algorithm and average slope method, is used to eliminate the feature pairs with mismatch. Thirdly, the classification of binocular vision, the parallax ranging method, the classification of binocular vision, the parallax ranging method, are used in this paper. The image feature point tracking method and robot motion parameter estimation method are introduced in detail, and the binocular vision localization method is verified experimentally. Then, the robot path planning method based on artificial potential field is studied. This paper introduces the realization of the traditional artificial potential field method and its shortcomings. In view of the problem of local minima in the traditional artificial potential field method, an improved algorithm combining the repulsion direction with the autonomous establishment of the virtual target traction point is proposed. Finally, the work done in this paper is summarized, and the future work is prospected.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP242

【参考文献】