自主移动机器人导航与定位性能测试研究

CORC > 沈阳自动化研究所 > 中国科学院沈阳自动化研究所 > 机器人学研究室

题名	自主移动机器人导航与定位性能测试研究
作者	周维
学位类别	硕士
答辩日期	2015-05-26
授予单位	中国科学院沈阳自动化研究所
授予地点	中国科学院沈阳自动化研究所
导师	李斌
关键词	导航性能测试轨迹分析导航性能指标导航测试设备
其他题名	Research on the testing of an autonomous robot's navigation and positioning performance
学位专业	控制工程
中文摘要	随着科技的日新月异，机器人发展愈加迅猛。无论是在极地科考、地震救援、外星探索，或者是军事侦察作战、废墟救援，亦或者是工业自动化、民用智能车、助老助残等等方面，机器人的作用越加明显。然而正因为机器人种类五花八门，发展方向各不相同，因此对机器人的相互比较与评价显得更加困难。而对机器人进行较为全面的纵横向比较能更容易发现机器人自身的优缺点，对未来机器人的发展提供一定的指导方向，这对机器人未来发展的作用意义重大。自主移动机器人作为机器人研究中的主流，一般用来独立完成各种不同任务，作为载体的移动机器人虽各不相同，但从本质技术上亦有许多相似之处。本文依托于“中国科学院科研装备研制移动机器人环境感知能力测试系统项目”，对自主移动机器人导航与定位性能测试部分做出研究。本文研究目的旨在初步为自主移动机器人导航性能评价提出统一的指标与方法，为以后建立真正统一具有普适性的机器人导航性能评估体系迈出尝试性的一步。针对于此，本文的主要研究内容如下：首先对国内外目前关于自主移动机器人导航性能测试的方法和指标体系等做了研究分析，关于此方面的方法比较散乱，各个研究人员的侧重点也各不相同。总的来说，目前较为主流的方法有模拟地形法、综合任务法、轨迹跟踪法等，其中模拟地形法和综合任务法测试需要搭建相应测试场景进行，轨迹跟踪法则对测试设备和计算能力要求较高。本文最终选取类似轨迹跟踪法对自主移动机器人导航性能测试进行研究，这样的优点是对测试的地形、地点没有太高要求，对机器人导航性能测试设备的使用性大大增强。接着介绍了目前主流的导航方法，包括惯性导航、视觉导航、卫星导航和其他一些导航方法，分析了各个导航方法的优缺点和适用场景。由于本文是基于高精度差分GPS研制的导航评测设备，因此继续介绍了GPS的基本定位原理、误差来源和差分GPS定位原理。机器人轨迹复杂多变，导航评测设备所采集数据散乱无章，为从散乱数据中提取机器人轨迹的整体性能，引进了李德毅院士的不确定性云模型。李德毅院士提出的不确定性云模型是从离散随机定量数据到定性整体属性转换的优秀工具，根据一维逆向云发生器，能从散乱正态分布的数据中提出云模型的三个数字特征期望、熵和超熵Cloud(Ex，En，He)，以此概括整条轨迹的优劣程度。然后参考国内外大量文献，提出了一套导航性能评价指标体系，包括速度指标、导航效率和定位精度三方面，由“最大速度 ”、“最小速度 ”、“平均速度 ”、“曲率能量 ”、“轨迹总长 ”、“所用时间 ”、 “最大误差 ”、“平均误差 ”、“误差期望 ”、“误差熵 ”和“误差超熵 ”十一项具体指标组成。并设计了静态单点定位测试方法和动态轨迹跟踪测试方法。动态轨迹跟踪有圆形路径、直线路径、正弦曲线路径和四边形路径，这能体现机器人在固定曲率、零曲率、变曲率及转弯闭环回路等各种轨迹下的自主导航性能，最后还加入了针对任意曲线路径的处理研究。最后根据以上提出的种种理论，设计了一套自主移动机器人导航性能评测设备，硬件上包括基站和移动站，软件上由在线数据采集和离线分析处理两部分组成。然后设计实验首先验证该设备已达设计精度要求，再利用该设备对我所研制的南极科考机器人和合肥物质科学研究院的无人车进行了导航性能测试，验证了该设备的有效性。
索取号	TP242/Z77/2015
英文摘要	With the progress of science and technology, robot development is even growing rapidly. Whether in the polar expedition, earthquake relief, extraterrestrial exploration, or military reconnaissance operations, rubble and rescue, or by industrial automation, civil smart car, HelpAge disabled, etc., the role of robots have become increasingly obvious. However because of robot variety and different development direction, it is more difficult for mutual comparison and evaluation of the robot. But the robot more comprehensive aspect to compare robot itself can be easier to find advantages and disadvantages, to provide some direction for the future development of robots, and it’s important for the future of robots. Autonomous mobile robot as a robot research in the mainstream, generally used to independently complete a variety of different tasks, as a carrier of a mobile robot, although different, but there are many similarities between the technical natures. In this paper, a preliminary study aimed to evaluate the performance of autonomous mobile robot navigation indicators and propose a unified approach for the future establishment of a truly unified with a universal robot navigation performance evaluation system tentative step forward. For this, this article main research content is as follows: First, I did a research and analysis for the current domestic and international methods and indicators on autonomous mobile robot navigation performance testing, and it’s very different for each other. Overall, the more mainstream methods are simulated terrain, comprehensive task, trajectory tracking, etc. . Simulated terrain and comprehensive task need corresponding test scenarios, and trajectory tracking needs test equipment and computing power. So we choose trajectory tracking finally. Then I introduced the current mainstream navigation methods, including inertial navigation, visual navigation, satellite navigation and other navigation methods, and analyzed the advantages and disadvantages of various navigation methods and applicable scenarios for them. This article is based on high precision differential GPS navigation metering equipment, so I continue to introduce the fundamental of GPS positioning principle, error sources and differential GPS positioning principle. The robot trajectory is complex and data collected is scattered, so I introduced the uncertainty cloud model form Academician Li Deyi. It can transform information from a discrete random data to quantitatively overall property. The uncertainty cloud model has three digital features, expectations, entropy and hyper entropy, Cloud(Ex，En，He). Then I referred to a large number of domestic and foreign literature, and put forward a set of navigation performance evaluation indicator system, including speed indicator, efficiency of navigation and positioning accuracy, which is composed of "maximum speed ", "minimum speed ", "average speed ", "curvature energy ", "track length ", "the time ", "maximum error ", "average error ", "error expected ", "error entropy " and "error hyper entropy " eleven specific indicators. After that, I designed the static single point positioning test method and dynamic trajectory tracking test method. The dynamic trajectory tracking includes circular path, straight path, sinusoidal path and quadrilateral path, I also added a research for arbitrary curve path finally. According to all of the above, I designed a set of autonomous mobile robot navigation performance evaluation equipment, the hardware includes a base station and a mobile station, the software analysis consists of online data acquisition and off-line processing two parts. Then I designed experiments to verify that the device had reached the design accuracy first. After that, I tested the navigation of an Antarctic expedition robot which was designed by SIA and an unmanned vehicles which came from Hefei Institutes of Physical Science, verified the effectiveness of the device.
语种	中文
产权排序	1
页码	67页
内容类型	学位论文
源URL	[http://ir.sia.ac.cn/handle/173321/16810]
专题	沈阳自动化研究所_机器人学研究室
推荐引用方式 GB/T 7714	周维. 自主移动机器人导航与定位性能测试研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2015.