基于IMM-UKF的协同式车辆运动状态跟踪算法

为了提高车辆运动状态的跟踪精度,在路侧传感网信息融合的基础上,提出了一种改进的车辆运动状态估计方法.采用匀速直线运动模型和匀速转向运动模型建立了多模型的汽车行驶状态方程,利用马尔科夫链进行模型切换.同时引入无迹卡尔曼滤波算法,根据前一时刻的运动状态和当前观测值,对车辆行驶的运动状态参数进行估计.结果表明,改进的多模型数据融合算法与单模型相比,轨迹和速度跟踪误差分别降低了86.8%和78.6%,有效地提高了车辆运动状态跟踪的精度.

Cooperative vehicle motion tracking algorithm based on IMM-UKF

In order to improve the tracking accuracy of vehicle motion states, an improved vehicle motion state estimation algorithm was proposed on the basis of information fusion of roadside sensor network. A multi-model vehicle running state equation was established using both uniform linear motion model and uniform steering motion model, and the model switch was accomplished with Markov chains. Meantime, an untracked Kalman filtering algorithm was introduced to estimate the motion parameters for running vehicles according to the previous and the current observation values of motion states. The results show that the track and velocity tracking errors of the improved multi-model data fusion algorithm reduce by 86.8% and 78.6%, respectively, compared with the single model, effectively improving the tracking accuraccy of vehicle motion states.