摘要近年来，无人驾驶汽车（简称“无人车”）在国民经济与国防安全等诸多领域具 有广阔应用，谷歌公司的无人车在过去几年的百万公里测试中仅发生十一起轻微事故。 因此，无人车自主导航的研究在道路交通安全等方面具有重要意义，本文重点研究多 传感器信息融合技术在无人车自主导航系统中的应用。83529

针对视觉传感器所采集信息易产生噪声和畸变这一现象，利用小波变换对无人车 视觉系统的原始数据做图像融合和预处理并分析了目标方位角的获取方法。同时，针 对超声波避障传感器的误差问题，采取 BP 神经网络的方法对距离测量值进行数据级 融合。最后，利用模糊逻辑信息融合建立起系统的决策级控制器，根据无人车运动学 二维参考坐标系，设计出无人车自主导航系统的仿真算法并获得仿真结果。

通过上述各类融合算法的 MATLAB 仿真实验，获取到较为清晰的小波变换融合 图像以及预处理后的边缘轮廓；构建的 BP 神经网络在数据级融合中满足目标误差， 拟合程度较高；用于自主导航系统决策级信息融合的模糊控制器仿真效果良好，生成 了直观的路径规划图。

毕业论文关键词：多传感器信息融合；小波变换；BP 神经网络；模糊控制

Abstract In recent years, driverless vehicle has broad prospects in national economic, security and many other fields, Google's driverless vehicle occur together only 11 minor accidents in the past few years。 Therefore, the study of autonomous navigation of driverless vehicles is of great significance in terms of traffic safety。 the paper focuses on multi-sensor information fusion technology in unmanned autonomous vehicle navigation system。

Owing to the phenomenon that vision sensor for the collection of information is easy to produce noise and distortion, this article use wavelet to transform raw data from unmanned vehicle vision systems for image fusion and Pretreatment, and analyzes the target azimuth。 Meanwhile, the paper discusses the error problem in ultrasonic obstacle avoidance sensor, and use BP neural network for measurement data level fusion。 Eventually, we use fuzzy logic information fusion to create the system controller。 according to the analysis of the kinematic model of an unmanned vehicle, we design the simulation algorithm for unmanned autonomous vehicle navigation system and achieve the simulation result。

Through the MATLAB simulation experiments about the above types of fusion algorithm, we obtain a clear image from wavelet transform and the fusion edge contour after pretreatment; the BP neural network we constructed meet the target level error in data fusion, and fit higher degree; the Simulink result of the Fuzzy controller for decision fusion in autonomous navigation system is good, and achieve the intuitive path plan。

Keywords: Multi-sensor Information Fusion; Wavelet transform; Fuzzy control; Neural networks

目 录

第一章 绪 论 1

1。1  引言 1

1。2  课题背景与意义 1

1。3  国内外研究现状 2

1。3。1  多传感器信息融合技术的发展现状 2

1。3。2  无人车自主导航的研究现状 2

1。4  课题的主要内容 3

第二章 多传感器信息融合技术 4

2。1  传感器的组成与分类