Research On A Simultaneous Localization And Mapping Method Aided By Magnetic Beacons

It is a challenging work to achieve accurate positioning for an autonomous vehicle without the use of global navigation satellite system(GNSS)and underwater acoustic positioning systems,especially in unobserved environments.Simultaneous localization and mapping(SLAM)technique has got much attention and research for its advantage in improving the vehicle’s positioning accuracy while observing the environmental characteristics.While the SLAM performance relies strongly on the richness of the natural environmental characteristics.To contribute a navigation method with stronger environmental adaptability,a magnetic beacon-aided SLAM method is studied in the thesis.The magnetic beacons can work either alone or with natural environmental characteristics to improve the positioning accuracy.This thesis focuses on the following three contents: simultaneous detection of multiple magnetic beacons,establishment of magnetic beacon-aided SLAM model,and the establishment of SLAM model that fuses magnetic beacons and multibeam bathymetric data.The main contributions are summarized as follows:1.A novel method that simultaneously localize multiple magnetic beacons is studied towards the use in navigation.A regularized objective function is constructed by taking the mixed magnetic field composed of magnetic beacons’ magnetic field and geomagnetic field as measurement.The Levenberg-Marquardt algorithm is taken to optimize the beacons’ position,magnetic moments as well as the background field.The number of beacons can be determined at the same time.The “high-wall effect”,which probably happens in the optimization process,is first proposed and avoided through reasonable initialization.Both simulation and field experiments are conducted to verify the effectiveness of the detection method.2.The SLAM models based on magnetic beacons are established.The SLAM models are established in both particle filter and graph-based frameworks to make performance comparison.By improving the data-association process,the potential misdetected magnetic beacons can be removed and the system robustness is enhanced.A field experiment is carried out in a small region using the MTI-G710 IMU and a group of magnetic beacons.The SLAM system performs well even with the strong background magnetic anomaly interference.The system also shows good fault recovery ability in the case that the vehicle suddenly loses its track for some reasons.3.The graph-based SLAM that fuses the magnetic beacons and multi-beam bathymetric data is established.In the front-end,the factors in the graph are obtained from multibeam trip matching,magnetic beacon detection and data association.In the backend,the Levenberg-Marquardt algorithm is taken to solve the positions of the vehicle and beacons.Thereby,the vehicle positioning,as well as the building of seabed bathymetric and beacon maps are achieved simultaneously.The multibeam bathymetric dataset measured in a region in Huanghai Sea is used in experiment.Results show that with the aiding of magnetic beacons,the SLAM system can get obviously higher accuracy compared with the case that uses the bathymetric data only.