为解决运梁车在过隧道时载梁片两侧距离隧道壁过窄的问题，提出一种基于粒子群算法的16线激光雷达实时检测隧道中心线的方法。对某型号运梁车，设计三维激光雷达支架，安装及标定激光雷达，构建隧道中心实时检测平台。使用粒子群算法和PCL（Point Cloud Library）中的点云滤波对激光雷达数据进行处理，实时求解出车辆前进方向相对于隧道纵向的横摆角度，在此基础上求出隧道壁横截面点云，并用粒子群算法拟合出该圆的中心点坐标和半径，从而拟合出隧道中心线。实验结果表明，该方法可有效检测出隧道中心线，同时该方法求出的隧道半径与隧道实际半径的误差以及提取的隧道中心线与实际中心线的误差均在3 cm以内，这表明文章提出的算法可以对隧道中心进行实时检测。

This paper proposed a real-time detection method for tunnel centerline by 16-line LiDAR based on particle swarm optimization （PSO） algorithm to solve too narrow of the beam''s two sides from the tunnel wall crossing the tunnel. We designed a three-dimensional LiDAR bracket for a certain type of beam truck， installed and calibrated the LiDAR， and constructed a real-time tunnel-center-detection platform for the beam transport vehicle. Using the PSO algorithm and point cloud filtering （PCL） in the point cloud library， we processed the LiDAR data and then calculated the real-time cross-swing angle of the vehicle''s forward direction relative to the vertical tunnel. Thus， we obtained the point cloud of the tunnel wall''s cross section and fitted the center point coordinate and radius of the circle using the PSO algorithm to fit the tunnel centerline. The experimental results show that the method can effectively detect the tunnel centerline. The errors between the tunnel radius calculated by this method and the actual tunnel radius and between the extracted tunnel centerline and actual tunnel centerline were both within 3 cm. Hence， the proposed method can effectively detect real-time tunnel centerline.