基于体素栅格的矿区敞车车厢自动定位与分割的研究

Research on Automatic Positioning and Segmentation of Open Wagon Carriage in Mining Area Based on Voxel Grid

在矿料车间的自动化改造过程中常采用多个安装在车间顶部的激光雷达来感知外部环境信息。然而,这种安装方式常导致雷达到火车的距离与矿料车间中敞车车厢之间的距离相差过大,雷达采集到的点云十分杂乱稀疏且部分缺失;随着车厢中堆积的矿料越来越多,雷达采集到的点云也会不断变化;且车间中的火车有不同的型号尺寸、在装完一部分车厢之后火车还会向前运动。这些情况都会给车厢的自动定位和分割带来困难。基于此,借鉴点云体素的思想,创新性地提出了一种基于平面矩形栅格的矿区敞车车厢的点云自动定位与分割方法,通过实际数据测得该方法的相对误差依然可以控制在±1%以内,具有良好的优越性能,可以为矿区的自动化改造提供先决条件。

In the process of automation transformation of the mineral material workshop,multiple laser lidars installed at the top of the workshop are often used to perceive external environmental information.However,this installation method often leads to a large difference between the distance from the radar to the train and the distance between the gondola cars in the mineral workshop,and the point cloud collected by the radar is very messy and sparse and partially missing.As more and more mineral materials are accumulated in the carriage,the point cloud collected by the radar will also change continuously.And the trains in the workshop have different models and sizes,and the trains will move forward after loading part of the carriage.These situations will bring difficulties to the automatic positioning and segmentation of the carriage.Based on this,drawing on the idea of point cloud voxels,an innovative point cloud automatic positioning and segmentation method for open wagon carriage in mining area based on planar rectangular grids was proposed.The relative error of the method measured by the actual data can still be controlled within±1%,which has good superior performance and can provide prerequisites for the automation transformation of the mining area.