CORC  > 清华大学
微型直升机自主飞行辅助视觉系统研究
那盟 ; 贾培发 ; NA Meng ; JIA Pei-fa
2010-06-09 ; 2010-06-09
关键词视觉系统 目标识别 位置估计 微型直升机 vision system,object recognition,position estimation,micro helicopter TP391.41
其他题名A Vision System for Autonomous Flight of Micro Helicopter
中文摘要为了实现微型直升机自主飞行,论文设计了实时的计算机视觉系统。该系统包括机载上的硬件设备和软件算法,能够实现目标识别、特征点提取和位置估计。文中设计了特定图标,并主要依据颜色来进行快速识别。此外,通过检测4个共面的图像特征点,能够在单目单帧的基础上计算出直升机相对于特定目标的三维位置信息。视觉系统任务由位于机载上的IntelSitsang板来执行。由于受到板子运算能力的限制,每秒钟可以处理5帧160×120的彩色图像。飞行实验结果表明,该视觉系统的目标识别率高达93%,角点计算偏差小于3个像素,同时自身定位的三维位置坐标平均偏差为8cm、13cm和6cm。因此,该视觉系统通过精确自身定位,能够很好的辅助微型直升机完成自主悬停、起飞和降落等飞行任务。; In order to achieve autonomous flight of micro helicopter,a real-time computer vision system is presented in this paper.This vision system consists of on-the-board hardware and customized software algorithm which implements object recognition,feature point extraction,and position estimation.A special target is designed and the recognition algorithm mainly depends on the color information which significantly simplifies and accelerates the vision tasks. Besides,the position parameters can be calculated under the condition of one camera and one frame,through extracting four coplanar feature points.The vision algorithms are processed by Intel Sitsang board on board.Because of the ability of Sitsang,it can process 5-frame images of 160 by 120 in a second.The results of the flight experiments show that the object recognition rate of the vision system reaches 93%,the extracted corner pixels are accurate to within 3 pixels,and the average errors of the position parameters are 8cm,13cm and 6cm.So,the vision system proposed in this paper can commendably assist micro helicopter to achieve autonomous hovering,launching,landing and other flight tasks.
语种中文 ; 中文
内容类型期刊论文
源URL[http://hdl.handle.net/123456789/55214]  
专题清华大学
推荐引用方式
GB/T 7714
那盟,贾培发,NA Meng,等. 微型直升机自主飞行辅助视觉系统研究[J],2010, 2010.
APA 那盟,贾培发,NA Meng,&JIA Pei-fa.(2010).微型直升机自主飞行辅助视觉系统研究..
MLA 那盟,et al."微型直升机自主飞行辅助视觉系统研究".(2010).
个性服务
查看访问统计
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。


©版权所有 ©2017 CSpace - Powered by CSpace