基于点云直方图的回环检测算法和车辆定位方法

Loop-closure detection algorithm based on point cloud histogram and vehicle positioning method

针对智能车辆在城市复杂路段由于全球定位系统GPS信号丢失导致的定位失准问题,建立因子图优化模型对激光雷达和惯性测量单元进行数据融合,提出一种紧耦合框架下的激光-惯性里程计(LIO)车辆定位方法,实时估计车辆状态信息;并提出基于点云直方图的回环检测算法,通过计算车辆当前位置与历史时刻位置间点云的相似度判断车辆是否到达同一位置,进而结合上一次经过该位置时的信息校正车辆当前状态,减少定位误差的积累。KITTI数据集上的测试结果表明:回环检测模块可有效降低LIO的误差积累,带有回环检测模块的LIO具备良好的定位精度。

Aiming at the problem of inaccurate positioning of intelligent vehicles due to the loss of GPS signals on complex urban roads,a factor graph optimization model is established to perform data fusion on LiDAR(Light detection and ranging)and Inertial Measurement Unit(IMU),and the lidar-inertial odometry(LiDARinertial odometry,LIO)vehicle positioning method under the tightly-coupled framework is proposed,which can estimate vehicle state information in real time;a loop-closure detection algorithm based on point cloud histograms is proposed.The similarity of the point cloud determines whether the vehicle has reached the same position,and then combines the information from the last time that the vehicle passed the position to correct the current state of the vehicle,reducing the accumulation of positioning errors.The test results on the KITTI dataset show that the loop-closure detection module can effectively reduce the error accumulation of the LIO,and the LIO with the loop-closure detection module has excellent positioning accuracy.