自动落布车位姿估计的准确性是影响其在纺织车间内同时定位与地图构建(simultane-ous localization and mapping,SLAM)的关键因素.在进行自动落布车位姿估计时,遇到观测噪声异常变化或噪声协方差与算法不匹配等情况时,无迹卡尔曼滤波(unscented Kalman filter,UKF)难以准确估计小车的位置和姿态.针对此问题,将误差序列协方差估计与遗忘因子同时引入U K F进行改进,提出了一种改进的自适应UKF自动落布车位姿估计算法.通过误差序列协方差估计对观测噪声协方差矩阵R进行调整,引入遗忘因子对R进行自适应更新,进而得到自动落布车位姿的最优估计.实验结果表明,在高斯噪声环境下,改进的UKF算法比其他算法具有更好的鲁棒性和估计精度.改进后的UKF位姿估计算法代入Cartographer算法后建图误差值减小,表明此算法能够在室内复杂环境下达到更加精确的位姿估计.
Automatic cloth dropping vehicle position estimation based on improved UKF
The accuracy of the position estimation of the automatic cloth dropping vehicle is a key factor affecting its simultaneous localization and mapping(SLAM)in the textile workshop.When estimating the position and state of the automatic cloth dropping vehicle,it is difficult to accu-rately estimate the position and state of the trolley by unscented Kalman filter(UKF)when encountering abnormal changes in the observation noise or mismatch between the noise covariance and the algorithm.In response to this issue,an improved adaptive UKF position and state esti-mation algorithm was proposed based on the UKF by introducing the error sequence covariance estimation and the forgetting factor into the UKF at the same time.Firstly,the observation noise covariance matrix R was adjusted by the error sequence covariance estimation,the forgetting fac-tor was introduced to adaptively update R,and then the optimal estimation of the position and state was obtained.The experimental results show that the improved UKF algorithm has better robustness and estimation accuracy than other algorithms in Gaussian noise environment.The improved UKF position estimation algorithm was substituted into Cartographer algorithm to build a better map,which further indicates that this algorithm is able to achieve more accurate position estimation in indoor complex environments.
国家科技期刊平台
NSTL
万方数据
