基于迭代无迹卡尔曼滤波的SLAM算法仿真研究

对迭代无迹卡尔曼滤波算法在SLAM问题中的应用进行仿真研究。通过仿真分析发现,与一般的无迹卡尔曼滤波算法相比,迭代的算法有时无法提高SLAM的精度,继而探讨了SLAM问题中选择采用迭代算法的条件;同时针对迭代算法的观测更新阶段,用阻尼的高斯-牛顿迭代方法改进完全高斯-牛顿迭代方法,从而提出一种改进的基于迭代无迹卡尔曼滤波的SLAM算法。仿真实验对提出的迭代条件进行了验证,仿真结果表明提出的SLAM算法与无迹卡尔曼滤波算法相比,可以进一步提高SLAM问题的估计精度。

Simulation Research of SLAM Algorithm Based on Iterated Unscented Kalman Filter

The SLAM algorithm based on iterated unscented Kalman filter was analyzed through simulation.Simulation results indicate that,comparing with the SLAM algorithm based on unscented Kalman filter,the iterated one could not improve the SLAM accuracy in some circumstances.To deal with this problem,the condition of using iteration in SLAM was discussed first;the damped Gauss-Newton iteration was adopted to replace the traditional full Gauss-Newton iteration during the observation update process of SLAM algorithm.A new improved SLAM algorithm based on interated unscented Kanlman filter was proposed.The simulation experiments validate the proposed iteration condition.The experiment results show that the proposed algorithm can further improve the SLAM accuracy.