基于鲁棒自适应SCKF的智能汽车目标状态跟踪研究

准确的自车和前车状态估计是智能汽车有效决策和控制的前提，而以往的研究通常不考虑噪声统计特性不确定的问题，导致某些情况下车辆状态估计的误差很大。为此，提出一种鲁棒自适应平方根容积卡尔曼滤波（Robust adaptive square-root cubature Kalman filter，RASCKF）算法，以降低噪声统计不确定性对估计精度的影响。首先，采用最大后验概率准则估计了过程噪声协方差和测量噪声协方差的统计值，以提高噪声稳定时状态估计的精确性。然后，基于标准化测量新息序列设计了故障检测规则，利用实时测量新息对噪声协方差进行校正处理，保证状态估计算法的鲁棒性。最后，在不同的噪声干扰工况下对RASCKF算法进行了仿真验证。结果表明，RASCKF算法在估计精度和稳定性上明显优于标准SCKF算法，有效地解决了智能汽车目标状态跟踪过程中噪声统计特性不确定的问题。

Research on Intelligent Vehicle Target State Tracking Based on Robust Adaptive SCKF

Accurate estimation of the states of vehicle and front vehicle is the premise of the effective decision-making and control of the intelligent vehicle. However, previous studies usually do not consider the uncertainty of noise statistical characteristics, which leads to a large error of vehicle state estimation in some cases. Therefore, a robust adaptive square-root cubature Kalman filter (RASCKF) algorithm is proposed to reduce the influence of noise statistical uncertainty on estimation accuracy. Firstly, the statistical values of process noise covariance and measurement noise covariance are estimated by the maximum a posterior (MAP) criterion to improve the accuracy of state estimation when the noise is stable. Secondly, the fault detection rules are designed based on the standardized measurement innovation sequence, and the real-time measurement innovation is used to correct the noise covariances for ensure the robustness of the state estimation algorithm. Finally, the RASCKF algorithm is verified by simulation under different noise interference conditions. The results show that RASCKF algorithm is superior to standard SCKF algorithm in estimation accuracy and stability, which effectively solves the problem of uncertain noise statistical characteristics in the process of intelligent vehicle target state tracking.