一种基于ST-EKF的惯性/里程计滑动窗口滤波技术

A SINS/odometer Sliding Window Kalman Filtering Technique Based on ST-EKF

针对在卫星拒止的复杂野外环境下传统的惯性/里程计组合导航系统存在的误差随时间积累和方差不一致的问题,提出了一种基于状态变换Kalman滤波(State Transformation Extended Kalman Filter, ST-EKF)的惯性/里程计滑动窗口滤波技术。一方面,引入滤波鲁棒性更好的ST-EKF替代扩展Kalman滤波(Extended Kalman Filter, EKF);另一方面,利用滑动窗口中储存的误差状态向量、误差协方差和状态转移矩阵等数据实时对车辆行驶的航迹进行修正,以达到提高导航定位精度的目的。总里程489.88km的长行驶里程激光陀螺惯性测量单元(Inertial Measurement Unit, IMU)车载导航实验数据的事后处理表明,该算法具有较好的航迹修正效果和较高的导航定位精度,相比于基于EKF的惯性/里程计组合导航算法,其半程(去程)水平位置的均方根误差(Root Mean Square Error, RMSE)减少了42.79%,全程水平位置的RMSE减少47.70%。

Aiming at the problem of error accumulation and variance inconsistency in the traditional inertial/odometer integrated navigation system with times in the complex field environment with satellite failure, an SINS/odometer sliding window filtering technology based on the state transformation extended Kalman filter(ST-EKF) is proposed in this paper. On the one hand, a more robust ST-EKF is used to replace the extended Kalman filter(EKF). On the other hand, by using the data of system model stored in the sliding window, such as state error vector, error covariance and state transition matrix, the track of vehicles is modified in real time to improve the navigation and positioning accuracy. Simulation by post-processing the long-endurance ring laser gyroscope inertial measurement unit(IMU) experiment data with a total distance of 489.88 km shows that this algorithm has a better track correction effect and high positioning accuracy. Compared with the integrated navigation algorithm of SINS/odometer based on EKF, the RMSE of the horizontal position in the half-way(outward voyage) is reduced by 42.79%, and that of the whole journey is reduced by 47.70%.