摘要
车辆位置的精确、可靠获取,一直是阻碍智能驾驶技术的难题。特别当车辆处于复杂道路环境中时,车辆卫星定位信号易受较大干扰,使车辆定位产生漂移现象。针对车辆定位的这种漂移现象,研究了针对车辆位置跟踪的卡尔曼-高斯联合滤波方法。对于车辆卫星定位受到的干扰不同,采用分层处理的滤波方法;针对卡尔曼滤波不能较好地滤除一些干扰较大的位置漂移点,通过设置与车速、航向角等相关的动态阈值,对卫星定位的车辆位置进行动态阈值判断;通过动态阈值识别出的车辆位置漂移数据,结合高斯过程回归,以车辆的历史数据作为学习样本,使用预测值和真实观测值构建补偿量,通过对卡尔曼观测方程加入动态观测补偿实现车辆位置优化;对于一般噪声产生的卫星定位波动,联合滤波也可以有效优化。实车实验表明,该方法可以有效识别出车辆定位的漂移点,车辆卫星定位在信号受较大干扰的情况下,车辆卫星定位的精度可以提高30%左右,最大误差由9 m降低到0.8 m左右。该联合滤波方法在使用低成本定位装置的情况下,有效提高车辆卫星定位的精度及可靠性。
Accurate and reliable acquisition of vehicle locations has always been a problem that hinders intelligent driving technology. Particularly when the vehicle is in a complex driving environment,vehicle’s satellite positioning signal is susceptible to significant interference,causing the vehicle’s positioning to drift. A joint Kalman-Gauss filtering model is developed to solve the drift in vehicle positioning. For different interference,the layered filtering method is adopted. For some position drift points with large noise that cannot be well filtered out by Kalman filtering,the accuracy of vehicle positioning is judged by setting dynamic threshold related to vehicle speed and heading angle,etc. The vehicle position drift data identified by dynamic thresholds is combined with Gaussian Process Regression,the historical data of the vehicle is used as a learning sample. The difference of predicted value and the actual position of the vehicle position is used to construct the compensation value. The vehicle positioning is optimized by adding dynamic observation compensation to the Kalman observation equation. For satellite positioning fluctuations generated by general noise,joint filter is also used. Real vehicle simulations show that this method can effectively identify the drift point of vehicle positioning. In the case of signal interference,the accuracy of satellite positioning can be improved by about30%. and the maximum error is reduced from 9 m to about 0.8 m. The joint filtering method can effectively improve the accuracy and reliability of vehicle satellite positioning when using low-cost positioning devices.
作者
高策
褚端峰
何书贤
贺宜
吴超仲
陆丽萍
GAO Ce;CHU Duanfeng;HE Shuxian;HE Yi;WU Chaozhong;LU Liping(Intelligent Transport Systems Research Center,Wuhan University of Technology,Wuhan 430063,China;Engineering Research Center for Transportation Safety of Ministry of Education,Wuhan University of technology,Wuhan 430063,China;School of Computer Science and Technology,Wuhan University of Technology,Wuhan 430063,China)
出处
《交通信息与安全》
CSCD
北大核心
2020年第1期76-83,共8页
Journal of Transport Information and Safety
基金
国家重点研发计划项目(2018YFB0105004)
国家自然科学基金项目(51675390、U1764262)资助。
关键词
智能交通
车辆位置跟踪
联合滤波
高斯过程回归
卡尔曼滤波
intelligent transportation
vehicle position tracking
joint filtering
Gaussian Process Regression
Kalman filtering