旋翼飞行机器人视觉定位方法及系统

我是walle

旋翼飞行机器人视觉定位方法及系统

旋翼飞行机器人视觉定位方法及系统.pdf (886.56 KB, 下载次数: 5)

2016-3-11 18:46 上传

点击文件名下载附件
下载积分: 金钱 -1

旋翼飞行机器人视觉定位方法及系统

李力，王耀南，刘洪剑，谭建豪，钟杭

（湖南大学电气与信息工程学院，湖南长沙410082）

摘要：为实现旋翼飞行机器人在室内环境中的精确定位及控制，设计了基于单目视觉的机载视觉定位方法．该方法通过矩特征提取、图形分类来识别地面标记图形，进而利用EPnP 方法解算摄像机坐标．为验证机载视觉定位方法的有效性，利用Kinect 传感器搭建了旋翼飞行器外部定位系统，设计了针对旋翼飞行器目标的视觉跟踪算法，定位精度达2 mm．实验结果表明，机载视觉定位方法能够很好地进行飞行器定位．
关键词：旋翼飞行机器人；视觉定位；矩特征；bagging 分类器；n 点透视
中图分类号：TP24 文献标识码：A 文章编号：1002-0446(2016)-01-0008-09

Visual Positioning Method and System for Rotor Flying Robot

LI Li，WANG Yaonan，LIU Hongjian，TAN Jianhao，ZHONG Hang

(College of Electrical and Information Engineering, Hunan University, Changsha 410082, China)

Abstract: To achieve exact positioning and control of the rotor flying robot in indoor environments, an onboard visual positioning (OVP) method based on monocular vision is designed. It can recognize landmark images by moment feature extraction and image classification, and calculate camera coordinates through EPnP method. To verify the effectiveness of the proposed onboard visual positioning method, an external visual positioning system (EPVS) based on Kinect is built, and a visual tracking algorithm for rotorcraft is designed. The experimental results suggest that the positioning deviation of the
rotorcraft is within 2 mm, and the OVP method is suitable for rotorcraft positioning.
Keywords: rotor flying robot; visual positioning; moment feature; bagging classifier; perspective n point