Algorithm Research And System Construction For Dynamic Target Detection And Tracking Under Autonomous Navigation Of Unmanned Vehicles

Unmanned Surface Vehicles(USV)are small surface vessels that can navigate at high speeds on the water surface,mainly used to perform tasks that are not suitable for manned vessels.In the autonomous navigation of unmanned boats,marine radar undertakes unknown target detection and tracking tasks,and plays an indispensable and important role in unmanned navigation.Marine radar requires signal processing in a very complex signal background to achieve the goal of detecting targets and maximizing the detection of true target signals.At the same time,through data processing,the point track and track processing of radar targets is further completed,thereby achieving highprecision real-time tracking of the target.The research topic of this thesis is algorithm research and system construction for dynamic target detection and tracking under autonomous navigation of unmanned vehicles.The main research content is the target detection algorithm and multi-target tracking algorithm based on the marine radar of the unmanned surface vehicles.At the same time,based on the research of algorithms,design and develop a marine radar target tracking software system.The main contents are as follows:(1)For the processing and receiving part of marine radar data,the main work is to receive and analyze the echo data from marine radar and the position information from inertial navigation system through the network communication module,and transform coordinate conversion according to the polar coordinate system conversion Cartesian coordinate system formula to obtain marine radar images and complete the construction of radar images.(2)For the characteristics of marine radar images,several common image segmentation algorithms are compared.Through theoretical research and comparison,the Density Based Spatial Clustering of Applications with Noise(DBSCAN)algorithm,which is most suitable for target detection in marine radar images,is selected for target detection.Aiming at the problem of high time complexity of DBSCAN algorithm,an optimization method based on distance search is proposed,which improves the computational speed of the DBSCAN algorithm.Experiments have shown that the improved algorithm greatly improves its computational speed in the case of multiple data points.To address the problem of a large number of pseudo targets in the DBSCAN algorithm’s segmented targets,an evaluation function method based on target morphology information is proposed for target filter,in order to filter out false targets and reduce the number of targets passed into the tracking algorithm.Experiments show that the proposed method has a good performance in recognition and filtering of false targets in marine radar images.(3)For the problem of target tracking,a comparison is made between the multi target tracking algorithm based on target detection and the multi target tracking algorithm based on Joint Probability Data Association(JPDA)and Kalman filter(KF).Experiments have shown that the JPDA-KF multi target tracking algorithm has excellent tracking performance.In response to the problem of high computational complexity in the JPDA data association algorithm when tracking a large number of targets,combined with the evaluation function method based on target morphology information proposed in chapter 3 of this thesis,an improved multi-target tracking scheme for marine radar based on target filtering is proposed.Experiments show that the algorithm processing scheme proposed in this thesis can effectively reduce the impact of false targets in marine radar images,and greatly improve the MOTA index of multi-target tracking algorithm.(4)Design and develop software system for marine radar target tracking.The technical core of the entire software system is based on the relevant algorithms studied in this thesis;the overall architecture of the software system is separated from the front and back ends,the front end is developed using Qt Quick/QML,and the back end is developed using C++,which ensures that the overall design of the software system conforms to modularization,the principles of scalability and maintainability.In the process of software system design and development,first carry out separate development and testing of each part and functional module,and then integrate all parts and functional modules into a complete software system,and finally complete the development of the software system,reach the initial design indicators,meet the actual use needs,and put into application in specific USV projects,and provides a stable multi-target detection and tracking solution for autonomous navigation of unmanned vehicles.