The invention discloses an automobile trajectory tracking method based on variational Bayesian particle filtering. Firstly, particles are initialized; for the problem of unknown system noise distribution, IW distribution is selected as conjugate prior distribution of a state prediction error covariance # imgabs0 #; then, a coordinate rising method is used for conducting iteration updating on the # imgabs1 #; the updated # imgabs2 # is substituted into Gaussian distribution to serve as an importance density function of the system, new particles are obtained through sampling from the function, and then weights are calculated and normalized; finally, resampling the particle set, wherein polynomial resampling is selected as a resampling method; an optimal estimation value is obtained, and trajectory tracking of the automobile is achieved. According to the method, the problems of difficulty in accurate assignment of parameters due to insufficiency of prior modeling information and low prediction precision of vehicle driving track estimation due to unknown system noise in existing practical engineering are solved.
本发明公开了基于变分贝叶斯粒子滤波的汽车轨迹跟踪方法,首先初始化粒子;针对系统噪声分布未知问题,选取IW分布作为状态预测误差协方差#imgabs0#的共轭先验分布;然后利用坐标上升方法分别对#imgabs1#进行迭代更新;将更新后的#imgabs2#带入高斯分布,作为系统的重要性密度函数,从中抽样得到新粒子,再计算权值,并归一化;最后对粒子集进行重采样,重采样方法选取多项式重采样;得到最优估计值,实现对汽车的轨迹跟踪。本发明解决了现有实际工程中,由于先验建模信息的不足导致难以对其参数进行准确赋值以及系统噪声未知导致估计汽车行驶轨迹预测精度低的问题。
Automobile trajectory tracking method based on variational Bayesian particle filtering
基于变分贝叶斯粒子滤波的汽车轨迹跟踪方法
18.02.2025
Patent
Elektronische Ressource
Chinesisch
AUTONOMOUS AUTOMOBILE GUIDANCE AND TRAJECTORY-TRACKING
Europäisches Patentamt | 2019
|Autonomous automobile guidance and trajectory-tracking
Europäisches Patentamt | 2022
|Kernel-based Bayesian filtering for object tracking
IEEE | 2005
|Automobile trajectory tracking control method based on rapid model prediction
Europäisches Patentamt | 2020
|