基于低成本激光传感器移动机器人SLAM研究与实现

发布时间：2018-03-16 19:41

本文选题：SLAM　切入点：扩展卡尔曼　出处：《山东大学》2017年硕士论文　论文类型：学位论文

【摘要】：机器人技术自问世以来,就得到了世界各国的重视,尤其是对智能移动机器人SLAM的研究,一直以来都是机器人技术学科的研究热点,从1986年SLAM诞生至今已有三十年的发展,目前研究SLAM的基本框架基本已经趋于稳定,而且已经有相当多的应用,但是研究的方法大都是基于高成本高精度的传感器,导致不能够实现产业化,随着近几年的传感器技术的发展,低成本的传感器应运而生,为移动机器人SLAM实现产业化提供了可能性,本文就是在这种情况下,研究与实现基于低成本的传感器移动机器人SLAM,主要研究的内容如下:1)分析了 SLAM的分类以及研究方法,并对比了各种研究SLAM的方法,同时根据SLAM的基本框架,构建了基于低成本的传感器的EKF-SLAM模型,使用卡尔曼理论解决移动机器人位姿问题,采用简单的码盘和陀螺仪数据建立的数学模型,模型越简单系统误差越小,系统的复杂性越低,越能接近实际,仿真与实际测试结果表明本文构建的数学模型的有效性。2)针对SLAM中地图构建的问题,本文采用低成本的激光传感器实现ICP-SLAM,实现了经典ICP算法以及PLICP,对比两者之间实验结果,同时为了防止算法失效本文采用EKF辅助的方法,当激光点匹配率低于60%的时候就代表该次匹配失效,为了防止缺失该时刻的地图,采用EKF在此时刻的估计位姿作为ICP算法的旋转R和平移T矩阵,由此机器人行走不至于很快的缺失环境信息,实验结果证明了提出的方法的有效性。3)为了提高精度,本文提出另一种方法,基于模糊理论信任度的二次数据融合,融合航迹推算位姿,EKF估计位姿以及ICP配准位姿,使用一种模糊隶属度函数描述对三者位姿的信任程度,通过计算三者的信任度矩阵,以及对应的权值,最终做出决策对EKF和ICP的信任程度。
[Abstract]:Since the birth of robot technology, it has been paid attention to by many countries all over the world, especially the research on intelligent mobile robot (SLAM), which has been the research hotspot of robotics science all the time. It has been 30 years since SLAM was born in 1986. At present, the basic framework of SLAM has become stable, and has been widely used. However, most of the research methods are based on high-cost and high-precision sensors, which lead to the inability to realize industrialization. With the development of sensor technology in recent years, low-cost sensors emerge as the times require, which provides the possibility for the industrialization of mobile robot SLAM. This paper studies and implements the sensor mobile robot slam based on low cost. The main research contents are as follows: 1) the classification and research methods of SLAM are analyzed, and the methods of studying SLAM are compared. At the same time, according to the basic framework of SLAM, the classification and research methods of SLAM are compared. The EKF-SLAM model based on low cost sensor is constructed, and the position and pose problem of mobile robot is solved by Kalman theory. The simpler the model is, the smaller the system error is, and the more simple the model is, the smaller the system error is. The lower the complexity of the system, the closer it is to reality. The simulation and actual test results show that the mathematical model constructed in this paper is effective. 2) aiming at the problem of map construction in SLAM, In this paper, the ICP-SLAM is implemented with a low cost laser sensor, the classical ICP algorithm and the PLICP algorithm are implemented, and the experimental results between the two algorithms are compared. In order to prevent the algorithm from failing, this paper uses the EKF aided method. When the laser point matching rate is lower than 60%, it represents the failure of the matching. In order to avoid missing the map of the time, the estimated position and orientation of EKF at this time are used as the rotation R and the translation T matrix of the ICP algorithm. The experimental results show that the proposed method is effective. 3) in order to improve the accuracy, another method is proposed in this paper, which is based on the fuzzy theory trust degree of quadratic data fusion. Fusion track estimation EKF estimation and ICP registration pose, a fuzzy membership function is used to describe the degree of trust of the three postures. The trust matrix and the corresponding weights of the three positions are calculated. Finally make a decision about the degree of trust in EKF and ICP.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】