【作者】 周金祥；

【导师】 李小坚；

【作者基本信息】 北方工业大学 ， 检测技术与自动化装置， 2006， 硕士

【摘要】 多传感器数据融合技术和自主移动机器人技术均为当前国际研究的热点,但是在我国尚处于发展初期。多传感器融合技术已经在“973”计划中作为鼓励研究领域重点推出,该技术在军事决策、工业控制、特种机器人、海洋监视和综合导航等领域有着广泛的应用前景。本文从理论和实践两方面对多传感器数据融合技术进行了卓有成效的探索,做出了一定实验的结果,主要包括以下四个方面的内容: 第一,介绍了多传感器数据融合技术在国内外发展的概况,以及多传感器数据融合系统的功能和结构模型,列举出了该技术在机器人方面成功运用的案例;同时介绍了几种常用的数据融合算法,如Bayes方法、D-S证据推理方法、模糊集理论、神经网络法等。 第二,将强跟踪滤波器(strong tracking filter)理论引入到多传感数据融合系统中。文中列出了该理论的关键推导步骤和相关定理,以及主要参数的计算公式。然后采用Matlab的m文件编程实现这一组递推算法。并对一个假设的典型多传感器融合非线性时变系统的状态和参数进行联合估计仿真,与扩展卡尔曼滤波器算法比较,得到了非常理想的效果,从而也丰富了数据融合理论。 第三,在自主移动机器人实验平台的软件框架中加入电子罗盘模块,并结合编码器、超声和红外传感器的信息,成功实现机器人定点方向调整与自主导航,同时也得出了这样的结论:只依靠电子罗盘导航是不精确的。编程中的难点包括整个程序框架和接口函数的理解、电子罗盘类的封装、串行口通信、线程间同步和通信等。 第四,自主设计和实现了基于AT89C52单片机的多红外传感器障碍物信息检测与环境温度检测系统。文中详细地给出了该系统的硬件设计和软件实现的全过程,在软件方面,在Keil C51环境下采用纯C语言编程实现。同时,在PC机上用Visual C++6.0编写了一个数据帧接收程序,完成了PC机与单片机间的通信。更多还原

【Abstract】 The technology of multisensor data fusion and autonomous mobile robot both are pop topics in the world at the present. But, in domestic relevant science fields, the technology is still in initial stages. Fortunately, the technology of multisensor data fusion was brought forth as an important research field in the Project 973. There is a broad application prospect about the technique in various fields, such as military decision-making, industry control, special robot, ocean scouting, and integrative navigation, etc. In this article, the author has performed fruitfully in theoretic aspect and practice aspect, and has gotten some significant results as well. There are four primary aspects about this article as below.Firstly, a worldwide survey of multisensor data fusion technology is presented, and the data fusion systems’ function and framework are described clearly. Some successful solutions about this technology used in robotic field are enumerated in this article. Meanwhile, several common algorithms about data fusion are introduced respectively, such as Bayes-method, Dempster-Shafer evidential reasoning, fuzzy set theory, neural networks techniques, and so on.Secondly, the theory of strong tracking filter is adopted in multisensor data fusion system. In the article, most key deducing steps and related theorems are recounted and primary parameters formulas are recounted as well. Furthermore, the author implements the algorithm by coding m-file in Matlab software environment. And the author uses this algorithm to calculate the states and parameters in a typical nonlinear data fusion system. According to the computer simulation, we can get that this algorithm is prior to the extended Kalman filter theory. It enriches the data fusion theory in some degree.Thirdly, an electronic compass module is installed in the autonomous mobile robot system. The author combines the data from electronic compass, coding sensors, infrasonic sensors and infrared sensors to make the robot adjust orientation accurately and navigate by itself. At the same time, the author draws such a conclusion: navigation only depending electronic compass is not enough and inaccurate. The difficulties during programming include understanding thewhole software framework and interface functions, encapsulating electronic compass class, serial port communication, multithread communication & synchronization, etc.Fourthly, the author designs a system based on MCU AT89C52, which can detect obstacles with infrared sensors and measure the environmental temperature with temperature sensor. The global procedure about the system’s hardware design and software implementation is described at length. In aspect of the software, the programs are coded in C-language in Keil C51 software environment. A program which receives data frame on PC is written in visual C++ language, and it completes the communication between PC and MCU.更多还原