SIA OpenIR  > 机器人学研究室
基于扩展集员滤波的空地机器人协同环境建模方法
Alternative TitleExtended Set-Membership Filter-based Air-ground Robots Cooperative Environment Mapping Method Research
杜文强1,2
Department机器人学研究室
Thesis Advisor谷丰
Keyword有界噪声 扩展集员滤波 路径规划 信息融合 协同环境建模
Pages72页
Degree Discipline模式识别与智能系统
Degree Name硕士
2019-05-17
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract随着技术的发展,地面自主移动机器人系统的发展有一个明显特点:机器人的工作范围正在从结构化、静态环境向非结构化、动态环境扩展,在科学考察,野外设施职守,军事侦察和边境巡逻等领域正逐渐获得越来越多的关注。因此,如何在广域野外非结构化的环境中有效感知周围的环境信息是移动机器人的研究热点。地面移动机器人对广域野外非结构化环境的感知主要是依靠自身的传感器,如相机、激光雷达等,对周围环境进行三维环境建模。2.5维地图环境模型是环境建模常用的环境描述方法,但是在野外环境中,由于地面机器人移动过程中存在的振动以及传感器的误差,造成地面移动机器人所构建地图的高程出现偏差,传统的概率估计方法无法实现高程范围的准确估计。针对这一问题,本文提出了一种基于边界鲁棒的地面机器人环境建模方法,首先构建传感器测量模型并获取测量误差的边界描述,在此基础之上,利用集员估计方法实现高程边界的估计,与传统的3σ界相比,可以获取高程的可靠范围,提升环境模型的鲁棒性和可靠性。然而,随着科技的发展,地面移动机器人总是因为感知能力不足而受限于复杂的环境,如负障碍物和超高障碍物环境。相比于仅能执行单任务、造价昂贵、鲁棒性不高的单体平台,多机器人协同环境感知系统不仅具备效率优势、数量优势,还能进行多任务、自主协同等作业。为了解决单体机器人感知能力不足的问题,本文提出与拥有大场景环境感知能力的无人机进行协同建模的解决方案。为了解决无人机和无人车的感知数据融合这一难点问题,提出了基于扩展集员滤波的2.5维高程图协同建模方法。该方法设计了一种空地数据快速融合算法,通过直接同时更新无人车和无人机的感知数据的方法取代以往两个观测数据集相交进行融合的方法,具有更高的实时性。为了验证所提具有鲁棒误差边界的相对高程环境地图的准确性,利用两种不同环境建模方法的结果进行对比分析。针对两种典型的环境:负障碍物和超高障碍物环境,设计相关实验来验证所提空地数据融合方法的有效性和可行性。并基于鲁棒误差边界的相对高程地图模型对周围环境进行了可通行性分析和路径规划,验证所构建地图在路径规划中的重要作用。
Other AbstractWith the development of technology, a distinct feature of the development of ground mobile robot systems is that their workspace is expanding from a structured, static environment to a unstructured, dynamic environment. In scientific investigations, field facilities, military reconnaissance and border patrols, etc. The field is gradually gaining more and more attention. Therefore, how to effectively identify the surrounding environmental information in the unstructured environment of the wide field is a research hotspot of mobile robots. The ground mobile robot's perception of the unstructured environment in the wide-area field mainly relies on its own sensors, such as cameras and laser radars, to map the surrounding three-dimensional environment. The 2.5D environment map is a common environment description method for environment mapping. However, in the wild environment, due to the vibration and sensor errors in the movement of the ground robot, the elevation of the grid map constructed by the ground mobile robot is deviated, and the traditional probability estimation method cannot achieve an accurate estimation of the elevation range. Aiming at this problem, this paper proposes a boundary-robust-based ground robot environment modeling method. Firstly, the sensor measurement model is constructed and the boundary description of measurement error is obtained. On this basis, the height estimation boundary is realized by the set member estimation method. It is estimated that compared with the traditional 3σ boundary, the reliability range of the elevation can be obtained, and the robustness and reliability of the environmental map can be improved. However, ground mobile robots are always limited by complex environments due to insufficient sensing capabilities, such as negative obstacles and ultra-high obstacle environments. Compared with a single platform that can only perform single tasks, costly, and low robustness, the multi-robot collaborative environment-perception system not only has efficiency advantages, quantitative advantages, but also multi-task, independent coordination and other operations. In order to solve the problem of insufficient sensing ability of single robot, this paper proposes a solution for collaborative modeling with UAVs with large scene environment awareness. In order to solve the key problem of sensory data fusion between UAV and UGV, a 2.5D elevation map collaborative modeling method based on extended set filter is proposed. A fast fusion algorithm for air-ground data is designed. It replaces the previous two observation data sets and merges them by directly updating the sensory data of UGV and UAV. It has higher real-time performance. In order to verify the accuracy of the proposed relative elevation environment map with robust error boundaries, the results of two different environmental mapping methods were used for comparative analysis. For two typical environments: negative obstacles and ultra-high obstacle environments, relevant experiments were designed to verify the effectiveness and feasibility of the proposed air-ground data fusion method. The relative elevation map based on robust error boundary is used to analyze the traversability and plan path, and then verify the important role of the constructed map in path planning.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/25179
Collection机器人学研究室
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院大学
Recommended Citation
GB/T 7714
杜文强. 基于扩展集员滤波的空地机器人协同环境建模方法[D]. 沈阳. 中国科学院沈阳自动化研究所,2019.
Files in This Item:
File Name/Size DocType Version Access License
基于扩展集员滤波的空地机器人协同环境建模(6543KB)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
[杜文强]'s Articles
Baidu academic
Similar articles in Baidu academic
[杜文强]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[杜文强]'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.