摘要
为解决当前惯性传感器误差过大无法满足自动驾驶需求的问题,对惯性传感器的误差来源进行了分析,建立了精简后向传播神经网络来表征环境温度、机械安装误差、机械振动与电磁干扰因素对误差的影响模型。并利用该方法对惯性传感器误差进行了标定,测试结果表明,该标定方法可以减小惯性传感器输出误差,使惯性传感器在自动驾驶中的应用成为了可能。
In order to solve the problem that the rather large error of inertial sensor can’t match the requirement of autonomous driving,a detailed analysis is made about the sources of the error.Afterwards a back-propagation neural network is established to represent the influence of environment temperature,mechanical installation error,mechanical vibration and electromagnetic interference.Furthermore,calibration is performed based on this method.The test results show that this method can decrease the output error of inertial sensor then a solution to apply inertial sensor on autonomous driving is addressed.
作者
杨莉
李冬雪
王强
赵目龙
李海波
Yang Li;Li Dongxue;Wang Qiang;Zhao Mulong;Li Haibo(Intelligent and Connected Vehicle Development Institute,China FAW Corporation Limited,Changchun 130013;State Key Laboratory of Comprehensive Technology on Automobile Vibration and Noise&Safety Control,Changchun 130013;Business Unit Passive Safety and Sensorics China,Continental Automotive Corporation(LYG)Co.,Ltd.,Changchun Branch,Changchun 130000)
出处
《汽车文摘》
2021年第9期45-48,共4页
Automotive Digest
关键词
惯性传感器
自动驾驶
后向传播神经网络
误差标定
Inertialsensor
Autonomousdriving
Back-propagationneuralnetwork
Error calibration
