Nonlinear estimation problem is investigated in this paper. By extension of a linear H_∞estimation with corrector-predictor form to nonlinear cases, a new extended H_∞filter is proposed for time-varying discrete-tim...Nonlinear estimation problem is investigated in this paper. By extension of a linear H_∞estimation with corrector-predictor form to nonlinear cases, a new extended H_∞filter is proposed for time-varying discrete-time nonlinear systems. The new filter has a simple observer structure based on a local linearization model, and can be viewed as a general case of the extended Kalman filter (EKF). An example demonstrates that the new filter with a suitable-chosen prescribed H_∞bound performs better than the EKF.展开更多
针对移动机器人噪声模型不确定性导致定位算法鲁棒性弱、精度低的问题,提出一种基于奇异值分解(Singular Value Decomposition,SVD)的自适应无迹H_(∞)滤波定位算法。该算法利用无迹H_(∞)滤波融合多传感器数据估计移动机器人位姿,并通...针对移动机器人噪声模型不确定性导致定位算法鲁棒性弱、精度低的问题,提出一种基于奇异值分解(Singular Value Decomposition,SVD)的自适应无迹H_(∞)滤波定位算法。该算法利用无迹H_(∞)滤波融合多传感器数据估计移动机器人位姿,并通过自适应调节滤波器参数γ,提高了移动机器人的定位精度。同时为了提高算法的鲁棒性,采用SVD分解代替常规Cholesky分解,避免了误差协方差矩阵在数值迭代过程中出现负定的情况。实验结果表明:相较于扩展H_(∞)滤波和粒子滤波算法,基于SVD分解的自适应无迹H_(∞)滤波定位算法具有精度高、鲁棒性强的优势。展开更多
文摘Nonlinear estimation problem is investigated in this paper. By extension of a linear H_∞estimation with corrector-predictor form to nonlinear cases, a new extended H_∞filter is proposed for time-varying discrete-time nonlinear systems. The new filter has a simple observer structure based on a local linearization model, and can be viewed as a general case of the extended Kalman filter (EKF). An example demonstrates that the new filter with a suitable-chosen prescribed H_∞bound performs better than the EKF.
文摘针对移动机器人噪声模型不确定性导致定位算法鲁棒性弱、精度低的问题,提出一种基于奇异值分解(Singular Value Decomposition,SVD)的自适应无迹H_(∞)滤波定位算法。该算法利用无迹H_(∞)滤波融合多传感器数据估计移动机器人位姿,并通过自适应调节滤波器参数γ,提高了移动机器人的定位精度。同时为了提高算法的鲁棒性,采用SVD分解代替常规Cholesky分解,避免了误差协方差矩阵在数值迭代过程中出现负定的情况。实验结果表明:相较于扩展H_(∞)滤波和粒子滤波算法,基于SVD分解的自适应无迹H_(∞)滤波定位算法具有精度高、鲁棒性强的优势。
