基于激光雷达和视觉融合的室内导航方法设计

Design of indoor navigation method based on fusion of LiDAR and vision

提出了一种激光雷达(LiDAR)与视觉图像融合的室内导航算法,准确获取狭小空间内目标的角度和距离信息,并将其应用于智能小车实际导航中。LiDAR测量中引入边界距离差作为迭代阈值来探测目标物体的距离和角度信息;视觉图像中使用轮廓逼近算法提取图像中心像素坐标进行位置解算,得到物体中心相对偏转角和偏转方向。将LiDAR测得的角度与视觉解算测得的角度两部分信息进行加权融合得出实际偏转角,求得目标物体的准确位置信息,实验结果表明:融合算法测得角度最大误差仅有0.53°。将该算法应用到机器人系统学习套件典型智能小车上进行目标物抓取测试,成功率达到了96.25%,证明了此算法的有效性。

An indoor navigation algorithm that fuses LiDAR and visual images is proposed to accurately obtain the angle and distance information of the target in a small space,and it is applied to actual navigation of the smart car.In LiDAR measurement,the boundary distance difference is introduced as the iterative threshold to detect the distance and angle information of the target object.In visual image,the contour approximation algorithm is used to extract the image center pixel coordinates for position calculation,and the relative deflection angle and deflection direction of the object center are obtained.The angle measured by the LiDAR and the angle measured by the visual solution are weighted and fused to obtain the actual deflection angle,and the accurate position information of the target object is obtained.The experimental results show that the maximum error of the angle measured by the fusion algorithm is only 0.53°.The algorithm is applied to the robot system learning kit to carry out the target grabbing test,and the success rate reaches 96.25%,which proves the effectiveness of this algorithm.