基于贝叶斯估计的超声红外复合测距系统  

Ultrasonic infrared composite range system based on Bayesian estimation

作　　者：徐兵[1] 姜艳青 周志杰[2] 张玉玲 张邦成[1] 

机构地区：[1]长春工业大学机电工程学院,吉林长春130012 [2]第二炮兵工程大学控制工程系,陕西西安710025 [3]徐州重型机械有限公司零部件研究所,江苏徐州221004

出　　处：《解放军理工大学学报（自然科学版）》2013年第4期365-371,共7页Journal of PLA University of Science and Technology(Natural Science Edition)

基　　金：国家863计划资助项目(2008AA040205);国家自然科学基金资助项目(61004069)

摘　　要：针对单一传感器在复杂环境中测量数据可靠性差以及同一类型多传感器测距盲区无法补偿的问题,提出了一种基于贝叶斯估计的不同类型多传感器复合测距方法。该方法综合了超声与红外2种传感器的优点,利用红外传感器对超声传感器的测距盲区进行补偿,解决了单一类型传感器存在的测距盲区问题。对同一类型多个传感器同时获取的数据动态建立置信距离矩阵,应用椭圆曲线表示的支持程度关系矩阵确定各传感器测量结果中的有效值,并对其应用贝叶斯估计的方法进行数据融合,使测距准确度进一步提高。基于超声红外复合测距移动机器人的仿真实验表明,采用贝叶斯估计方法在保证实时性的前提下提高了测距的准确度,误差为±1mm,达到了设计要求,为移动机器人的研究提供了参考。Due to the fact that the data of single sensor measurement has poor reliability in complex environments,and multi-sensor of same type has ranging blind which can not be compensated,a composite distance measurement method with different types of multi-sensor was proposed based on Bayesian estimation.The proposed method combined the advantages of ultrasonic sensors and infrared sensors,used the infrared sensors to compensate the distance blind spots of the ultrasonic sensors,so that the issues of the ranging blind of single type sensor was solved.After establishing a dynamic confidence from matrix for the data obtained simultaneously from the multi-sensor of same type,the valid values were determined with the application of relationship matrix of the elliptic curve,which expresses supporting degree.Then the valid values were integrated by Bayesian estimation methods,which further improved the accuracy of the ranging.Measurement experiments were done through the developed ultrasonic infrared composite measure ranging mobile robot.Simulation and experimental results show that at the premise of real-time performance,the ranging accuracy was improved after adopting Bayesian estimation method.The error is ±1 mm,which meets the design requirements and provides reference for the study of robot.

关 键 词：复合测距 超声传感器 红外传感器 数据融合 贝叶斯估计 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]