【作者】 张鋆豪； 杨旭升； 冯远静； 李永强； 张文安；

【Author】 Junhao Zhang;Xusheng Yang;Yuanjing Feng;Yongqiang Li;Wen-an Zhang;College of Information Engineering,Zhejiang University of Technology;

【机构】 浙江工业大学信息工程学院；

【摘要】 本文研究了基于单目视觉的运动刚体位姿估计问题,提出了基于自适应无迹卡尔曼滤波算法(Adaptive Unscented Kalman Filter,AUKF)的位姿估计方法.考虑到运动刚体位姿估计系统的量测方程为非线性且过程噪声统计特征未知,通过递推噪声估计器在线估计过程噪声的均值和方差阵,解决了位姿估计系统中过程噪声统计特性未知时估计精度下降的问题.实验结果表明,AUKF算法提高了位姿估计的精度,并实现了过程噪声统计特性的在线估计.更多还原

【Abstract】 In this paper, the pose estimation of moving rigid body based on monocular vision is studied, an adaptive unscented Kalman filter(AUKF) based pose estimation method is proposed. The measurement equation of the moving rigid body pose estimation system is nonlinear, and the statistical characteristics of process noise cannot be accurately obtained in practice. By using the recursive noise estimator to estimate the mean and variance of the process noise online, the problem of estimation accuracy declination without knowing the statistical characteristics of the process noise can be overcome. The experimental results demonstrate that AUKF algorithm improves the accuracy of the pose estimation and realizes the online estimation of the statistical characteristics of the process noise.更多还原