多传感器融合的智能车辆导航研究  

Study on Navigation of Intelligent Vehicles Based on Multi-sensor Fusion

作　　者：王檀彬[1] 陈无畏[1] 焦俊[1] 汪明磊[1] 

机构地区：[1]合肥工业大学,合肥230009

出　　处：《中国机械工程》2009年第11期1381-1385,共5页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(70771036);清华大学汽车安全与节能国家重点实验室开放课题基金资助项目(KF2007-06)

摘　　要：为了解决单一传感器信息获取的不足,提出了一种融合视觉和超声波信息进行导航的方法,给出了从视觉信息中提取路面标志线及判断障碍物方位的算法。用D-S证据理论和模糊处理的方法将视觉分析得到的环境信息与多个超声波传感器获得的信息进行融合。算法包括了图像分割和数学形态学处理、超声系统的建模和信息的融合。仿真与实验结果表明:该方法可以有效提取出路面标志线并且信息处理和融合方法能有效判断障碍物,获得良好的避障导航效果。Aiming at the insufficiency of information acquired by a single sensor, a method of fusing the information acquired by camera and ultrasonic sensors was proposed for navigation. A method for segmenting the navigation path and identifying obstacle through the information of vision was put forward. D-S evidential theory and the method of fuzzy logic were used to fuse the information of the environment obtained by image processing and multiple ultrasonic sensors. Methods of information fusion, modeling of the ultrasonic sensors, image segmentation and mathematical morphology were included herein. The simulation and experiments indicate that the navigation path can be segmented efficiently and this method of information processing can identify obstacles efficiently and makes satisfactory results of navigation and obstacle avoidance.

关 键 词：信息融合 D-S证据理论 模糊逻辑 智能车辆 图像处理 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]