采用改进Sage-Husa自适应滤波的运动目标三维定位方法

提出不依赖于测距信息,利用两架基于视觉的无人机对运动目标进行三维交会定位的方法。采用多模型交互方法实现在不预知目标运动模式的条件下对运动目标的实时定位；采用改进的Sage-Husa自适应滤波算法,综合协方差匹配技术和正定性判断,提高了定位精度。为评估这些方法的性能,模拟真实观测条件进行仿真。结果表明,提出的方法可以实时对运动目标的三维坐标进行估计。改进的Sage-Husa自适应滤波算法可以显著提高定位精度,在90°观测夹角下,平均估计误差从27.13 m降低到14.62 m。仿真研究了两无人机观测夹角对定位的影响,结果表明:过小的夹角不利于定位精度的提高；较大的夹角对无滤波定位方法有较好的效果,但对基于改进的Sage-Husa自适应滤波算法的定位方法影响并不明显。

Three-dimensional localization for moving target using modified Sage-Husa adaptive filter

A three-dimensional intersection localization method for moving target using two vision-based UAVs (unmanned aerial vehicles), which did not rely on the distance information from the target point to the UAV was proposed. An interacting multiple model estimator was adopted to the localization method to solve the problem of not knowing the motion form of the moving target. A modified Sage-Husa adaptive filtering algorithm that synthesized the covariance matching technique and the positive definiteness judgment was used to improve the accuracy of localization. To assess the performance of these approaches, a set of simulations that carried out under realistic conditions were presented. Results show that the method proposed can get the accurate three-dimensional coordinates of the target. The modified Sage-Husa adaptive filtering algorithm can improve the localization accuracy significantly, with the average estimation error reduced from 27.13 m to 14.62 m under the intersection angle of 90°. The influence of the intersection angle on localization was studied in the simulation, which shows that too small intersection angle is not conductive to the improvement of localization accuracy, a larger intersection angle is good for the localization method without filtering, but the effect on the method with the modified Sage-Husa adaptive filtering algorithm is not significant.