Research On Airport Vehicle Passable Area Detection Method Based On Multi-Sensor Fusion

With the increasingly busy passenger and freight transport,it is more and more complex to keep the airport operating efficiently.As one of the solutions,the demand and acceptance of unmanned vehicles are gradually improving at domestic and foreign civil aviation airports.This paper studies the method of multi-sensor fusion of special vehicles in airport environment,builds multi-sensor experimental platform and designs and verifies the algorithm of obstacle detection and accessible area extraction,which lays a certain foundation for realizing the intelligent of special vehicles on the ground.Around the airport vehicle accessible area detection technology,this paper mainly makes the following research:Firstly,the method of extracting passable area based on laser point cloud is studied.On the basis of point cloud preprocessing and ground segmentation,the improved DBSCAN algorithm is used to cluster the non ground obstacles,and the obstacles are marked with different colors.For the ground points,the algorithm of distance between adjacent point clouds is used to extract the trafficable area of the ground points.Finally,the results of obstacle detection of non ground points and trafficable area extraction of the ground points are combined to obtain the final trafficable area extraction results based on the laser point cloud.Secondly,the obstacle detection method based on transfer learning model is studied.Aiming at the problem of small target of ground obstacle in airport scene,a transfer learning model is established based on the concept of transfer learning.With perception V3 as the source domain model and airport ground scene as the target domain,the airport ground obstacle detection based on vision is completed.Thirdly,a multi-sensor fusion model is constructed to realize simultaneous interpreting of different sensor data to monitoring environment.The internal parameter matrix of the camera and the external parameter matrix between the lidar and the camera are obtained to complete the multi-sensor spatial fusion,and the multi-sensor time fusion is completed based on the lidar time stamp to obtain the multi-sensor space-time fusion.Finally,the vehicle platform is used to detect the passable area,and 3D lidar and camera are used to detect the passable area.The experimental results show that for the problem of detecting the passable area of special vehicles on the ground,on the premise of realizing the obstacle detection and passable area extraction of a single sensor,calibrating two different sensors can complete the passable area of multi-sensor fusion.Compared with the passable area detection based on a single sensor,the method of multi-sensor fusion has higher accuracy The accuracy of the method is high.