This paper presents a novel experimental design to greatly improve the calibration accuracy of the acceleration-insensitive bias and the acceleration-sensitive bias of the dynamically tuned gyroscopes(DTGs).In order to reduce experimental cost,the D-optimal criteria with constraints are constructed.The turntable positions and the number of test points are chosen to build D-optimal experimental designs.The D-optimal experimental designs are tested by multi-position calibration experiment for tactical-grade DTGs.Test results show that,with the same cost,the fit uncertainty is reduced by about 50% by using the D-optimal 8-position experimental procedure,compared to using a defacto standard experimental procedure in ANSI/IEEE Std 813-1988.Furthermore,the new experimental procedure almost achieves optimal accuracy with only 12-position which is half the cost of the widely adopted 24-position experimental procedure for achieving optimal accuracy.
设计了一种无人机视觉/惯性组合导航系统,将无人机和地标点的运动模型作为状态方程,视觉信息作为观测量构建了与之对应的滤波模型.在滤波处理上,采用了复杂加性噪声模型对系统噪声进行建模处理;将小波分析引入到UKF（Unscented Kalman Filter）滤波中得到小波-UKF滤波算法,以此克服视觉观测噪声对滤波的影响;采用最大后验概率准则（MAP,Maximum A Posterior）自适应估计观测噪声协方差阵,并将其反馈到滤波过程中克服了小波处理后观测噪声方差阵不易确定的不足.仿真结果证明：对滤波算法的改进可以有效地提高滤波估计的精度.
