SIA OpenIR  > 机器人学研究室
城市环境下无人机控制与定位方法研究
Alternative TitleControl and Localization of Unmanned Aerial Vehicle under Urban Environment
代波
Department机器人学研究室
Thesis Advisor徐卫良 ; 何玉庆
Keyword城市环境 抗风扰控制 多传感器融合 优化与滤波 无人机
Pages112页
Degree Discipline机械电子工程
Degree Name博士
2020-05-27
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract本研究以城市环境中无人机的应用为背景,针对城市环境下急需解决的无人机抗风扰控制和弱或无 GPS 条件下的定位问题进行研究。主要展开以下几点研究内容:(1)基于加速度反馈增强的几何控制方法为了提高无人机的风扰抑制能力,实现无人机安全飞行和精准控制,提出了一种基于加速度反馈增强的几何控制抗风扰方法。该方法不需要改变原有的控制器结构,在原有的几何控制基础上引入角加速度和线加速度反馈,从而实现更快速且更高精度的角度和位置跟踪,提升系统的扰动抑制能力。同时,为了将该算法部署在实际的无人机平台上,提出了一种简单、快速以及适用于加速度反馈的无人机参数辨识方法。(2)加速度反馈增强的级联 H∞ 扰动抑制方法加速度反馈方法并不改变系统本身的稳定特性,但由于几何控制方法是一种非线性方法,且引入了积分器,系统针对风扰的稳定性较难以从理论上得到保证。因此在此基础上,提出基于加速度反馈增强的级联 H∞ 扰动抑制方法。首先将无人机的动力学解耦成内外环的形式,然后针对解耦后的系统设计一个级联 H∞ 控制器,最后引入加速度反馈增强方法到控制器中。系统整的抗扰动稳定性可以由 H∞ 设计的性能指标来保证。(3)基于优化的尺度不敏感多传感器融合方法利用视觉的多传感器融合可以提升无人机城市环境弱或无 GPS 条件下的定位精度。但是由于现有的视觉里程计方法在大尺度场景下会出现位置和尺度的漂移。针对该问题,提出基于图优化的尺度不敏感多传感器融合方法来融合局部视觉里程计信息和全局传感器如 GPS 和气压计等信息。通过将里程计的位置和尺度漂移描述成从里程计局部坐标系到全局坐标系间的相似变换,并优化包含多个最新状态序列的位姿图滑窗,来估计该变换,从而可以将局部视觉里程计实时地变换为全局位姿信息,得到全局无漂移局部高精度的状态估计。(4)结合滤波与优化的多传感器融合框架经过优化后的状态信息可能会由于某个异常点造成状态跳变。针对该问题,提出结合滤波与优化方法的多传感器融合框架。滤波为优化提供姿态约束,融合多个非同步传感器,以及平滑输出轨迹。同时在无人机受到风扰后,或者无人机为了应对风扰会产生较大的姿态变化,可能造成视觉信息特征丢失,里程计出现发散和错误。因此在该融合框架的基础上,提出基于卡方检测的传感器信息故障检测方法,从而提升系统在部分传感器出现故障时的鲁棒性。本文的研究工作包括了无人机在有挑战性的城市环境下风扰抑制控制和弱或无 GPS 条件下多传感器融合定位方法。通过实际的飞行控制和定位实验验证了所提出算法的有效性。本研究的方法使得无人机在城市环境中的飞行具有更高的控制和定位精度与鲁棒性。
Other AbstractWith the development of electronic manufacturing and information technology, smart city becomes the new goal of urban development to optimize management, improve efficiency and raise the life quality of citizens. Unmanned aerial vehicle (UAV) has the characteristics of vertical take off and landing, fixed point hovering and flexible maneuverability, and plays an increasingly important role in the construction of smart city, such as security and infrastructure inspection, road traffic monitoring and urban high precision map collection. However, UAV flying in urban environment has been suffering from many problems. It is very easy to emerge uncertain disturbance flow near buildings in urban environment, which makes the stability and high precision control of UAV extremely difficult. In addition, because of the building blocking or the electromagnetic interference, the GPS signal becomes weak, and the traditional localization method relying on GPS alone is no longer reliable. Therefore, based on the application of UAV in urban environment, this research focuses on the problem of UAV wind disturbance rejection control and localization in weak or no GPS environment. The main research contents are as follows: (1) Wind disturbance rejection based on acceleration enhanced geometric control To improve the ability of wind disturbance rejection, realize the safe flight and precise control of UAV, we propose a acceleration feedback (AF) enhanced geometric control method. This method doesn't need to change the original controller structure. Based on the original geometric controller, the angular and linear acceleration feedback are introduced to achieve faster and more accurate angle and position tracking and improve the disturbance rejection ability of the system. Meanwhile, in order to deploy the algorithm on the practical UAV platform, a simple, fast and applicable parameter identification method of UAV is proposed as well. (2) AF enhanced hierarchical H∞ method for wind disturbance rejection The AF method won't change the stability of the system against disturbance, but because the nonlinear characteristic of geometric control method and the introduction of integrator, it is difficult to guarantee the stability of the system against wind disturbance. Therefore, on this basis, a hierarchical H∞ disturbance rejection method based on AF theory is proposed. Firstly, the dynamics of UAV is decoupled into the form of inner and outer loop, then a hierarchical H∞ controller is designed for the decoupled system, and finally the AF enhanced method is introduced into the controller. The robustness against disturbance of the whole system can be guaranteed by the H∞ theory as well. (3) Scale insensitive multi sensor fusion method based on optimization Utilizing vision is a method to improve the localization accuracy of UAV in weak or no GPS environment. However, The existing visual odometry method suffers from position and scale drift. To this end, a scale insensitive multi sensor fusion method based on graph optimization to align local frame of odometry and global sensor information such as GPS and barometer is proposed. The position and scale drift is described as a similar transformation. By optimizing the slide window of pose graph which contains the latest states, the transformation can be estimated. So that the local visual odometry can be transformed into the global state in real time, and the globally drift free and locally accurate state estimation can be achieved. (4) Multi sensor fusion framework by combining filtering and optimization The optimized state may have jump due to some outliers in optimization progress. Thus, a multi sensor fusion framework combining filtering and optimization method is proposed. The filter provides attitude constraints for the optimization, fuses multiple asynchronous sensors, and smooths the output. Meanwhile, when the UAV is disturbed by the wind, or in order to reject wind disturbance, the UAV may have a big attitude change, which may cause a divergence of visual odometry because tracking loss. Therefore, based on this framework, a fault detection method of sensors based on Chi square tests is proposed to improve the robustness of the system. This research includes wind disturbance rejection control of UAV and multi sensor fusion localization in urban challenging environment. The effectiveness of the proposed algorithm is verified by the practical flight control and localization experiments. The proposed methods make UAV flying in urban environment more robustly and accurately both in control and localization aspects.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/27158
Collection机器人学研究室
Affiliation中国科学院沈阳自动化研究所
Recommended Citation
GB/T 7714
代波. 城市环境下无人机控制与定位方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2020.
Files in This Item:
File Name/Size DocType Version Access License
城市环境下无人机控制与定位方法研究.pd(18164KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[代波]'s Articles
Baidu academic
Similar articles in Baidu academic
[代波]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[代波]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

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