摘要
在复杂室内环境下,移动机器人一般采用多传感器融合的方法来实现精准定位。针对实际噪声和先验噪声统计特性不同而导致无迹卡尔曼滤波(UKF)算法精度急剧下降的问题,本文以双轮差速移动机器人为研究对象,提出了一种自适应UKF定位算法。首先,以UKF算法为基础,融合里程计、陀螺仪、激光雷达定位系统等传感器数据;然后,根据激光雷达定位系统的误差统计特性,预先校准观测噪声的均值和协方差矩阵,并采用自适应噪声估计器在线估计未知系统噪声的统计特性,来提高滤波的数值稳定性,减小状态估计误差;最后,使用搭载激光雷达R2000的双轮差速车MIR100进行实验,并与UKF算法的估计结果进行对比。实验结果表明:自适应UKF定位算法具有较强的鲁棒性,能够在复杂室内环境下实现较高精度的位姿估计。
In complex indoor environment, mobile robots generally use multi-sensor fusion method to achieve accurate positioning.Aiming at the problem that the precision of unscented Kalman filtering(UKF)algorithm decreases sharply due to different statistical characteristics of actual noise and prior noise, two-wheel differential mobile robot is taken as research object and an adaptive UKF localization algorithm is proposed.Firstly, based on the UKF algorithm, the sensor data such as odometer, gyroscope and LiDAR positioning system are fused.Then, according to the error statistical characteristics of LiDAR positioning system, the mean and covariance matrix of the observation noise are calibrated in advance, and the adaptive noise estimator is used to estimate the statistical characteristics of the unknown system noise online, so as to improve the numerical stability of filtering and reduce the state estimation error.Finally, the two-wheel differential MIR100 equipped with LiDAR R2000 is used for experiment, and the estimation results are compared with those of UKF algorithm.The experimental results show that the adaptive UKF localization algorithm has strong robustness and can achieve high-precision pose estimation in complex indoor environment.
作者
许万
朱力
张宇豪
方德浩
XU Wan;ZHU Li;ZHANG Yuhao;FANG Dehao(School of Mechanical Engineering,Hubei University of Technology,Wuhan 430068,China;School of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan 430070,China)
出处
《传感器与微系统》
CSCD
北大核心
2022年第11期44-47,56,共5页
Transducer and Microsystem Technologies
基金
船舶振动噪声重点实验室基金资助项目(6142204200709)。
