一种基于深度特征的室外环境下激光地图辅助视觉定位方法

针对无GPS或弱GPS信号下的室外环境中的车辆无法定位问题,提出了一种利用激光地图辅助视觉定位方法。首先利用双目相机的视差图的深度与三维激光雷达地图进行匹配,然后通过最小化深度残差来估计六自由度相机位姿,接着利用视觉跟踪产生的良好的初始估计和提出的深度残差方法可有效地估计相机的位姿,最终通过估计相机的位姿完成定位。通过对比多个公开数据集,验证所提方法的准确性和有效性,最后利用实验小车采集校园数据,仿真和实验结果都证明利用此方法的有效性和在室外环境下的视觉定位的准确性。

A Laser Map-aided Visual Location Method in Outdoor Environment Based on Depth Characteristics

Aiming at the problem of vehicle location in outdoor environment without GPS or weak GPS signal, a laser map aided visual positioning method is proposed. Firstly, the depth of disparity map of binocular camera is matched with that of three-dimensional lidar map, then the position and attitude of six-DOF camera are estimated by minimizing the depth residuals, and then the position and attitude of the camera are effectively estimated by using the good initial estimation and the proposed depth residuals method generated by visual tracking, and finally by estimating the phase. The position and posture of the machine are positioned. By comparing several open data sets, the accuracy and validity of the proposed method are verified. Finally, campus data are collected by the experimental car. The simulation and experimental results prove the validity of this method and the accuracy of visual positioning in outdoor environment.