一种极坐标下基于高斯过程的扩展目标跟踪方法

对于使用雷达系统的扩展目标跟踪,通常将原始测量值从极坐标转换为笛卡尔坐标,然后馈入估计器.但其转换测量期望值在目标真实方位上存在偏差.另外,测量误差协方差的计算取决于实践中不可用的真实状态.针对这一不足,本文提出了一种在概率数据关联框架下基于高斯过程的扩展目标跟踪方法.该方法首先,直接在极坐标中对测量函数进行建模,并求解出测量噪声协方差;其次,建立扩展目标的联合跟踪门对测量进行筛选,并获得测量新息;最后,计算关联事件概率并估计扩展目标的状态.仿真结果表明了该方法的有效性.

A Gaussian process based extended target tracking method in polar coordinates

For the extended target tracking with radar system, the raw measurements are usually converted from polar to Cartesian coordinate and fed to the estimator. However, there is a bias in the expectation of the converted measurement along the bearing. In addition, the calculation of measurement error covariance depends on the true state that is unavailable in practice. In order to address this problem, a Gaussian process based extended target tracking method in the probabilistic data association framework is proposed. First, the measurement function is modeled directly in polar coordinate and the measurement noise covariance is given. Second, the validation gate of the extended target is established and the measurement innovation is obtained with the validated measurements. Last, the association event probability is calculated and the state of the extended target is estimated. Simulation results show the effectiveness of the proposed method.