摘要
在室内拒止条件下,无人地面车辆的室内导航面临巨大挑战。为了解决这个问题,我们引入了一种基于激光雷达(LiDAR)和惯性测量单元(IMU)的松组合导航系统,并提出了一种新的基于LiDAR的混合扫描匹配算法来估计车辆的相对位置和方向变化,该算法结合了遍历搜索和数学优化算法。首先在遍历搜索算法中引入了适应度函数,以寻找车辆姿态最优估计值,然后将该估计值作为数值优化算法的初始值进一步优化,最后通过无迹卡尔曼滤波器(UKF)将LiDAR估计位姿与DRS预测的结果进行松组合,从而减少惯性传感器的累积误差。在自主开发的无人地面车辆平台上进行的实际实验结果表明,本文所提出的室内无人车导航系统可以有效地减小DRS的误差累积,与hectorslam方法相比,定位误差降低了92.4%。
Indoor navigation of unmanned ground vehicles(UVS)is facing great challenges under indoor rejection conditions.To solve this problem,we introduce a loose integrated navigation system based on LiDAR(LiDAR)and inertial Measurement Unit(IMU),and propose a new liDAR-based hybrid scan matching algorithm to estimate the relative position and direction changes of the vehicle,which combines traversal search and mathematical optimization algorithm.First introduced in the traversal search algorithm,fitness function,to find the optimal estimate vehicle attitude,then the estimate as the initial value of numerical optimization algorithm is further optimized,finally,no trace kalman filter(UKF)will LiDAR estimate pose with loose combination forecasting results of DRS,and thus to minimize the cumulative error of the inertial sensors.A practical experiment is carried out on a self-developed unmanned ground vehicle platform.The results show that the proposed indoor unmanned vehicle navigation system can effectively reduce the error accumulation of DRS,and the positioning error is reduced by 92.4%compared with hectorSLAM method.
作者
祝瑞辉
蔚保国
李爽
ZHU Ruihui;YU Baoguo;LI Shuang(State Key Laboratory of Satellite Navigation System and Equipment Technology,Shijiazhuang 050081,China;The 54th Research Institute of CETC,Shijiazhuang 050081,China)
出处
《数字通信世界》
2022年第4期14-17,共4页
Digital Communication World
基金
“导航定位与5G通信多模一体化关键技术”项目的支持,该项目是河北省重大研发专项,合同编号为20310901D。
