CAJ | 학술논문

针对深水机器人导航定位算法的鲁棒性问题，提出基于M估计的扩展卡尔曼滤波算法。该方法通过采用M估计评价函数引入加权矩阵，从而对观测异常值进行修正。不仅将M估计应用到非线性系统，同时也很好地抑制了观测异常值对导航算法的影响。通过仿真实验验证了该方法的有效性。
침대심수궤기인도항정위산법적로봉성문제，제출기우M고계적확전잡이만려파산법。해방법통과채용M고계평개함수인입가권구진，종이대관측이상치진행수정。불부장M고계응용도비선성계통，동시야흔호지억제료관측이상치대도항산법적영향。통과방진실험험증료해방법적유효성。
To improve the robustness of navigation and positioning algorithm for deep water robot,we propose an M-estimate-basedextended Kalman filter (EKF)algorithm.By the use of M-estimate evaluation function,the method introduces weighting matrix so as tocorrect the observed outlier.This applies the M-estimate in nonlinear system,and well retrains the impact of the observed outlier onnavigation algorithm too.The effectiveness of the algorithm is verified through simulation test.