基于RANSAC的地下矿山巷道边线检测算法

Roadway Edge Detection Algorithm Based on RANSAC in Underground Mine

巷道边线是井下铲运机反应式导航系统中重要的感知信息。为了准确可靠地在井下环境中感知巷道边线,提出一种基于二维激光扫描信息和随机抽样一致性(RANSAC)的巷道边线检测算法。首先计算每个激光点的曲率,根据曲率阈值将激光点云划分为多个区域;然后基于RANSAC从每个区域提取直线,并根据铲运机航向角及巷道的设计标准进行筛选;最后合并筛选后的激光点云数据,使用RANSAC算法生成最终的巷道边线。基于地下矿山6种典型的巷道场景对算法效果进行验证,结果显示提取的巷道边线可靠度均达到96%以上,且算法具有很高的实时性和稳健性。

Because the working environment of underground LHD(load-haul-and-dump-machine)is very bad,and with the increase of mining depth in underground mines,the realization of underground unmanned LHD is of great significance for ensuring the safety and health of workers and improving the production efficiency of mining enterprises.Navigation and positioning of LHD is one of the difficulties in the research of unmanned LHD.At present,the navigation technology of underground LHD mainly includes plan-based metric navigation and reactive navigation.The reactive navigation technology has the advantages of low cost and low computation.The former reactive navigation technology mainly relies on adding beacons manually,it has the shortcomings of high cost and poor adaptability.The roadway edge is an important natural beacon perception information,which has natural advantages compared with the artificial beacon.Foreign scholars had applied it to the reactive navigation system of underground LHD and achieved good navigation effect.However,they only did the research on the detection of the roadway edge in the straight roadway,no further discussion on the detection of roadway edge in more complex underground environments.Therefore,a more applicable roadway edge detection algorithm is proposed in this paper.This method is based on two-dimensional laser scanning information and random sampling consistency(RANSAC).The flow chart of the algorithm is as follows:Firstly,the curvature of each laser point in the laser point cloud is calculated,according to the curvature threshold,the laser point cloud data are divided into several regions.RANSAC algorithm is used to extract the roadway edges from each region.Then,the roadway edges are filtered according to the heading angle of the LHD and the design criteria of the roadway.Lastly,the laser point cloud data contained in the remaining roadway edges is merged,and the final roadway edges is generated by RANSAC algorithm again.This article simulated the laser data of six underground mine roadway scenarios,and these six sets of data included typical scenes from simple to complex in underground mine.The experiment was based on MATLAB,and the experimental results were analyzed from the aspects of parallelism,proportion of interior points,fit degree with heading angle,visual display and so on.The calculated results show that the reliability of the extracted roadway edges is more than 96%,and the visual results are in line with the actual situation.This method can detect roadway edge in various scenarios of underground mines,and has high robustness,the roadway edge detection algorithm can play an important technical support role in the reactive navigation of underground LHD,and is of great significance to the realization of the unmanned underground LHD.