本文主要研究了基于视觉的地面移动机器人对接技术，主要包括对接系统的设计、颜色分割、目标识别、摄像机标定、单目测距和机器人控制策略等算法，在最后实现对接时整个系统表现较为良好。本文研究方向主要有：65849

      将PC机上置摄像机作为机器人对接时的“眼睛”，首先寻找到目标对接站的方向，再通过单目测距得到机器人与对接站之间的距离。PC机得到位置信息后再由此控制机器人向目标对接站处移动。

     在摄像机标定时主要依靠的是图像处理技术，主要步骤为：图像与处理，图像分割，图像边缘和重心的提取。其中图像分割采用基于HSI颜色阈值的分割策略对对接站的标记图像进行分割，当目标图像的颜色为单一色且同时能与背景颜色明显区分时取得了很好的分割效果；图像边缘提取使用简化的Canny算子；重心的提取则使用标记轮廓进行计算，而不使用传统的区域像素求取，减少运算时间。

   而在对接站的搜索和命令机器人移动的研究上，利用贴在白色墙壁上的彩色圆形实现目标识别算法，再通过机器人视野的划分（文中方法将其分为3个部分）根据摄像头标定的目的地调整机器人位置，最终实现对接。

毕业论文关键词 ：对接机器人，单目测距，HSI颜色分割，目标识别，Canny算子，椭圆拟合，控制策略

毕业设计说明书（论文）外文摘要

Abstract

This paper studies the ground mobile robot based on visual docking technology, including docking system design, color segmentation, object recognition, camera calibration, monocular robot control strategies ranging and algorithms to achieve docking in the final performance of the entire system is relatively well . This research directions are:

       Set the camera to the PC docking as the robot's "eyes", the first docking station to find the direction of the target, and then get through the monocular distance between the robot and the docking station distance. PC, then get location information thus controlling the robot is moved to the target docking station.

      The camera calibration relies mainly on image processing technology, the main steps are: image and processing, image segmentation, image edge and the center of gravity of the extract. Which HSI color image segmentation threshold based segmentation strategy docking station marker image segmentation, image color when the target is a single color and background color can be simultaneously obvious distinction achieved good segmentation results; image edge extraction using a simplified the Canny operator; gravity contour extraction is calculated using tags instead of using traditional regional pixels strike, reducing the computation time.

    While in the docking station search and command the robot to move research on the use of stickers on a white circle on the wall color to achieve target recognition algorithm, robot vision through the pision (the proposed method is pided into three parts) based camera calibration destination adjust robot position, and ultimately to achieve docking.

Key word: docking robot, monocular distance measuring, HSI color segmentation, object recognition, Canny operator, ellipse fitting, control strategy

目录

Abstract 2

1 绪论 2

1.1背景及意义 2

1.2发展动态 2

1.3研究内容 3

2 本文对接系统总体设计 5

2.1 对接系统的结构