本文为奥地利维也纳科技大学（作者：Dimitar Naydenov）的学士论文，共36页。

随着人们对自动驾驶汽车的日益关注，机器人自主导航已成为科学界的一个重要课题。为了在运动环境中导航，机器人首先需要对其当前位置进行估计，以便规划路径并相应地跟踪路径。有以下几种方法可以实现：其中一个例子是同时定位与映射（SLAM）技术，该技术对机器人周围的环境进行全面扫描，以便创建一幅具有机器人位置的障碍物地图。为了支持SLAM方法，本文开发了一种用于移动台坐标检测的位置跟踪系统。它依赖于将来自输入节气门和转向角的数据与来自两个独立传感器的测量值相结合，使用扩展卡尔曼滤波器（EKF）作为校正机制。传感器的测量依赖于使用惯性测量单元（IMU）检测移动基座的方向，并通过借助激光鼠标传感器估计机器人相对于其下方地面的位移来确定移动距离。该方法已经过实施和评估，在平面区域距离达10米的机器人坐标估计中，精度能够达到3%。

Autonomous robot navigation has becomeimportant for the scientific community due to the increasing interest inself-driving vehicles. In order to navigate in its environment, a robot needsto perform first an estimation of its current position, so that it could planits route and follow it accordingly. There are several approaches to accomplishthat: one example is the Simultaneous Localisation And Mapping (SLAM)technique, where a comprehensive scan of the surroundings of the robot isperformed, so that an obstacle map with the position of the robot in it couldbe created. In this thesis a position tracking system for detecting thecoordinates of a mobile station is developed to support the method of SLAM. Itrelies on combining data from the input throttle and steering angle withmeasurements from two separate sensors, using an Extended Kalman Filter (EKF)as a correction mechanism. The sensor measurements rely on detecting theorientation of the mobile base with an Inertial Measurement Unit (IMU) anddetermining the travelled distance by estimating the displacement of the robotrelative to the ground beneath it with the help of a laser mouse sensor. Thisapproach has been implemented and evaluated, achieving accuracy of 3% in theestimation of the robot’s coordinates for distances up to 10m on a flat surface.

1 引言

1.1 问题描述

1.2 相关方法

1.3 评估平台

2 实验设置

2.1 硬件配置

2.2 软件设计

3 结果

3.1 力学模型

3.2 传感器模型

3.3 扩展卡尔曼滤波器

4 总结

附录A 传感器测量

更多精彩文章请关注公众号：