Research On Lidar Target Detection And Tracking Method For Ship Intelligent Navigation

With the rapid development of computer,network communication,information processing and artificial intelligence,as well as the industry’s higher demand for ship efficiency,green and safety,intellectualization and unmanned have become the inevitable trend in the field of ship and marine engineering.As the premise and foundation of ship intelligent navigation,intelligent perception is one of the key technologies to realize ship intelligence.Water surface target detection and tracking are two main functions of intelligent sensing system.The former mainly processes the information collected by the sensor to obtain the basic information of the target of interest;The latter is based on the target detection results to monitor and update the target state in real time,so as to provide continuous and stable target state information for subsequent modules.Lidar is one of the commonly used sensors in ship intelligent sensing system.It has the advantages of high detection accuracy and rich target information.This paper studies the target detection and tracking method based on lidar for ship intelligent navigation.The main work includes:(1)Aiming at the interference of waves on point cloud data in the process of water surface target detection,a method of wave point cloud elimination in double density grid map is proposed.Through the strategy of constructing dense grid map in the small area where the target is located and sparse grid map in other places,this method can not only ensure that the wave point cloud is eliminated,but also reduce the memory occupation and computing time.Aiming at the problem that the traditional clustering algorithm is difficult to deal with the uneven distribution of point clouds,an improved DBSCAN(density based spatial clustering of applications with noise)clustering method is proposed in this paper.Based on the principle of clustering,the approximate distance between the laser radar and the reflection point and the vertical reflection point is calculated by using the algorithm.The experimental results show that the wave point cloud elimination algorithm proposed in this paper not only effectively removes the wave point cloud,but also improves the efficiency of the algorithm and reduces the calculation time;Compared with traditional methods,the improved DBSCAN algorithm has better clustering effect on point cloud targets with different densities and improves the performance of target detection.(2)Aiming at the interference of clutter and false alarm in the process of water surface single target tracking,after analyzing the causes of clutter and false alarm,a volume correction probability data association algorithm is proposed.The algorithm combines the geometric characteristics of point cloud targets and uses the volume ratio of point cloud after clustering to modify the calculation method of correlation probability in traditional methods,so as to improve the anti-interference ability of the algorithm.In addition,according to the requirements of ship encounter scenarios in international maritime rules,three experimental scenarios under classical encounter situations are designed to verify the superiority of the improved algorithm and the effectiveness of the algorithm in the process of ship collision avoidance.The experimental results show that the volume modified probabilistic data association algorithm proposed in this paper significantly improves the tracking accuracy and can effectively track the target in three classical encounter scenes.(3)Aiming at the correlation error caused by clutter,false alarm and track crossing in the process of water surface multi-target tracking,an improved km(Kuhn-munkres)data association algorithm is proposed.This method combines probabilistic data association with km algorithm,and integrates the interconnection probability into the similarity calculation of KM algorithm,so as to improve the robustness of the algorithm and reduce association errors.In addition,aiming at the interruption of target track caused by environmental interference and occlusion in the process of multi-target tracking,a track management method based on life cycle theory is introduced to increase the stability of track and reduce the interruption of track through multi threshold and multi-stage management mechanism.The experimental results show that the improved km algorithm proposed in this paper can effectively improve the robustness of the algorithm,and the track management method based on life cycle theory can significantly reduce the occurrence of track interruption.