中国科学院沈阳自动化研究所机构知识库
Advanced  
SIA OpenIR  > 机器人学研究室  > 学位论文
题名: 自动目标跟踪与无人机自主降落中的视觉方法研究
其他题名: The Visual Method Study in Automatic Object Tracking and UAV Autonomous Landing
作者: 范保杰
导师: 唐延东
分类号: TP391.41
关键词: 无人机视觉系统 ; 目标跟踪 ; 权重主动漂移纠正 ; 空间共线性误差
索取号: TP391.41/F23/2011
学位专业: 模式识别与智能系统
学位类别: 博士
答辩日期: 2011-11-29
授予单位: 中国科学院沈阳自动化研究所
学位授予地点: 中国科学院沈阳自动化研究所
作者部门: 机器人学研究室
中文摘要: 旋翼无人机具有独特的飞行性能和独有的低成本、低损耗、零伤亡、战场生存能力强、可重复使用以及高机动等诸多优势,在军用及民用方面具有广泛的应用前景。开展无人机自主环境感知技术的研究,在满足旋翼无人机在各领域应用的迫切技术需求,提高无人机自主能力等方面具有重大的意义。视觉系统作为旋翼无人直升机系统重要的组成部分,在实现环境感知和辅助导航方面具有得天独厚的优势。本文以实现无人机视觉导航为应用背景,针对其中的三个问题:自动目标跟踪,基于合作目标的无人机位姿估计,无人机自主降落中的平坦区域检测,进行了深入的研究,主要工作如下: 1.  针对移动目标跟踪中的模板漂移问题,提出了一种基于权重主动漂移纠正的模板跟踪算法。算法可以有效阻止跟踪过程中的模板漂移,无需再次进行纠正跟踪。总的能量函数包含两部分:跟踪项和漂移纠正项,通过最小化总的能量函数来同时实现模板跟踪和权重主动漂移纠正。主动漂移纠正项被纳入传统仿射配准算法的框架中。其最小化是通过基于权重L2准则的反向合成算法来实现的,另外,算法中的权重系数是自适应更新的。为了减少跟踪过程中的累积误差,我们还设计了一种新颖的模板更新策略,从先前模板序列中选择具有最小匹配误差的模板作为当前模板。不同场景的跟踪实验验证了我们算法的有效性,其性能全面优于Matthews[12]的被动模板跟踪算法。 2.      提出了一种基于合作目标的无人机的位姿估计方法。首先设计了两种合作目标,并发展了两种与之对应的实时图像预处理算法。接着提出了一种基于空间共线性误差的实时位姿估计算法,算法包含两个阶段:深度估计和绝对方位求解。通过最小化目标函数,迭代求解特征点的深度信息和位姿信息,接下来,利用绝对方位求解算法来求得最终的位姿信息。上述两个阶段依次迭代求解,直至结果收敛。仿真和实际实验表明,与当前的多种位姿估计算法相比,算法可直接应用于共面和非共面的情况,具有较高的精度,较好的鲁棒性。将该方法与两种合作目标结合,完成了模拟无人机自主降落的位姿估计实验,验证了方法的有效性。 3.      针对无人机自主降落中平坦区域检测问题,提出了基于序列图像的场景中平坦区域检测方法。首先针对序列图像的重建,提出了一种渐进式运动推断结构算法,对场景进行三维重建,获得场景中的三维点云信息,接下来利用最小中位数和Ransac算法从重建点云中找到近似场景地表的最佳平面,此最佳平面可以作为无人机的安全降落区域。在四种不同场景中,所提出的重建算法都获得了较稠密的三维点云,取得了令人满意的重建结果。平坦区域检测算法稳定准确的找到了近似当地地表的最佳平面,并与实际情况相符,验证了所提出的平坦区域检测方法的有效性。 4.      设计并开发了两种不同平台(PC104, DSP)的无人机视觉系统。该系统由机载视觉子系统、地面站子系统、无线通信子系统三个部分组成,构建了完整的空地、人机交互环路。室内室外的目标跟踪实验验证了所搭建系统的有效性,稳定性,可靠性。特别是基于DSP的无人机视觉系统,已完成多次飞行试验,取得了较好的实验效果。
英文摘要: Rotor UAV (Unmanned Aerial Vehicle) has excellent flight performance and unique low-cost, low-loss, zero injuries, and has battlefield survivability. It can re-use and has high mobility and many other advantages. For Rotor UAV, there is a wide range of applications in military and civilian aspects. The research on UAV sensing environment technology has an important significance to meet Rotor UAV the urgent application in various fields of technology needs, and to improve the UAV autonomy and other aspects. Visual system is as an important part of UAV system. It has unique advantage in achieving situational awareness and assisted navigation. In this paper, taking the visual navigation of UAV as research and application background, we focus our study on the three aspects: automatic object tracking, UAV autonomous landing based on cooperative targets, the safety zone detection in UAV autonomous landing. The work in this dissertation mainly includes: 1、For the template drift in moving object tracking, we propose a novel algorithm for object template tracking and its drift correction. It can prevent the tracking drift effectively, and save the time of an additional correction tracking. In our algorithm, the total energy function consists of two terms: the tracking term and the drift correction term. We minimize the total energy function synchronously for template tracking and weighted active drift correction. The minimization of the active drift correction term is achieved by the inverse compositional algorithm with a weighted L2 norm, which is incorporated into traditional affine image alignment (AIA) algorithm. Its weights can be adaptively updated for each template updating. For diminishing the accumulative error in tracking, we design a new template update strategy that chooses a new template with the lowest matching error. We will present various experimental results that validate our algorithm. These results also show that our algorithm achieves better performance than the Matthews’ passive drift correction algorithm[12]. 2、An UAV autonomous landing technology is presented. Firstly, we design two kinds of cooperative targets and develop two corresponding real-time image pre-processing algorithm. Secondly, we present an accurate and globally convergent pose estimation algorithm to solve the PNP problems based on space col-linearity error. The algorithm consists of two steps: the depth estimation and the absolute orientation. Based on the inverse projection ray, which starts from the optical center, and points to the object space through the image space, the object space col-linearity error is regarded as the cost function. Meanwhile, the principle depth and the relative depth of points are introduced to remove the residual errors of the cost function. By minimizing the cost function, we iteratively solve the pose information and depth of points respectively, and then reconstruct their coordinates in object space. In the following, the optimal absolute orientation solution gives the relative pose between the estimated 3D point set and the 3D mode point set. This procedure with the above two steps is repeated until the result converges. The proposed algorithm can be used directly in coplanar and non-coplanar point configurations. Our Experiments on simulated and real data indicate that the new algorithm has no less accurate than the state of art methods, and outperforms all tested methods in robustness to outliers and noise. Combining with the two cooperative objects, we completed the simulated UAV landing pose estimation experiment. The experimental results validate the effectiveness of our technology. 3、For the safety zone detection in UAV, we propose a scene flat region detection method based on the image sequence. Firstly, for the reconstruction of image sequence, we propose a progressive structure from motion algorithm to reconstruct the scene, and obtain the information of three-dimensional points in the scene. In the following, using the minimum median and Ransac algorithms to find the best plane near the scene surface from the reconstructive points, the best plane can be used as the safety landing zone for the UAV. In four different scenarios, the proposed reconstruction algorithm obtains dense 3D points, and achieves satisfactory results of the reconstruction. Then, the flat region detection algorithm finds the best plane approximation for local surface. These tests verify that the proposed method is effective. 4、Two UAV visual systems are designed and developed on different platforms (PC104, DSP). They include visual airborne subsystem, ground subsystem, wireless communication subsystem, a complete sky-land and human-computer interaction loop were constructed. Indoor and outdoor target tracking experiments validate the effectiveness, stability and reliability of the designed systems. Especially the visual system on the DSP platform has been used in  some real flight tests, and achieved good performance.
语种: 中文
产权排序: 1
内容类型: 学位论文
URI标识: http://ir.sia.cn/handle/173321/9412
Appears in Collections:机器人学研究室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
自动目标跟踪与无人机自主降落中的视觉方法研究.pdf(9945KB)----限制开放 联系获取全文

Recommended Citation:
范保杰.自动目标跟踪与无人机自主降落中的视觉方法研究.[博士学位论文].中国科学院沈阳自动化研究所.2011
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[范保杰]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[范保杰]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.

 

 

Valid XHTML 1.0!
Copyright © 2007-2016  中国科学院沈阳自动化研究所 - Feedback
Powered by CSpace