隧洞移动机器人里程计激光雷达融合定位

Mobile Robot Localization with Odometry and Lidar Fusion in Tunnel

在隧洞局部湿滑、不平整环境中,使用GPS对机器人进行定位会不准确甚至失效,仅使用里程计定位长时间累积误差会比较大。为了尽量减小机器人在这种环境中的定位误差,设计了一种定位方法,即先使用UMBmark算法离线校核机器人系统关键参数,且校核前机器人系统误差为0.3719 m,校核之后机器人的系统误差为0.2915 m,根据该算法定位精度评判依据,得出里程计定位精度提高了28%.然后利用扩展卡尔曼滤波算法,融合激光雷达和里程计进行SLAM,MATLAB仿真实验结果表明,融合前机器人x轴平均误差为2.047 m,y轴平均误差为1.245 m,航向平均误差为0.196 rad;在用EKF-SLAM融合激光雷达之后,机器人x轴平均误差为0.093 m,y轴平均误差为0.014 m,航向平均误差为0.003 rad。机器人平均位姿误差至少减小了一个数量级。仿真和实验结果证明了本文设计的提高隧洞局部环境定位精度方法的有效性。

In the local wet and slippery environment of the tunnel,using GPS for localization of the robot will be inaccurate or even invalid,The long-term cumulative error is relatively large only using the odometry.In order to minimize the localization error of the robot in this environment,designing a localization method,firstly using UMBmark algorithm to correct key parameters of the robot,the systematic error before correction is 0.3719 m,that is 0.2915 m after correction,and according to the localization accuracy evaluation of the algorithm.The localization accuracy has been improved by 28%.Then using the extended Kalman filter(EKF)algorithm,the lidar and odometry are fused to perform SLAM.Before fusion,average error ofxaxis is 2.047 m,y axis is 1.245 m,heading is 0.196 rad.And after fusion,average error ofxaxis is 0.093 m,y axis is 0.014 m,heading is 0.003 rad.The simulation results of MATLAB show that the average pose error of the robot is reduced at least by an order of magnitude after fusion.The simulation and experimental results prove the effectiveness of the proposed method for improving the local environmental localization accuracy of the tunnel.