Research On Localization Method Of Mobile Robot Based On Multi-sensor Information Fusion

Robot localization is the most basic and important task in the field of robotics research.Regardless of an intelligent autonomous mobile robot or a remotely controlled robot,it is necessary for robot's executing mechanism to consider the robot's pose and orientation.The localization method of outdoor environment usually uses GPS or GPS/INS integrated localization.However,the ordinary GPS's localization error is large,and the system will lose signal due to shelter in the tunnel,high buildings and woods.Localization of GPS/INS has characters of high localization accuracy in short term and cumulative errors in a long term.Therefore,INS can only smooth the localization data to a certain extent,and the upper limit of localization accuracy still depends on the localization accuracy of GPS.As a result,this thesis integrates the LIDAR-SLAM localization data based on the Cartographer,A graph optimization method,to achieve GPS/INS/LIDAR based on the integrated localization of GPS/INS integrated localization system.The work in this thesis is divided into four parts:1.The four coordinate systems commonly used in the field of integrated navigation and localization are studied and the conversion formulas between the coordinate systems are derived.In addition,the inertial navigation system,global positioning system,SLAM system based on lidar,and extended Kalman filter under the loose coupling framework are studied and related formulas are derived.2.The commonly used SLAM frameworks are: Hector,Gmapping,LOAM and Cartographer.Among them,Hector does not require odometer information but requires a high frequency scan of lidar.Gmapping relies on odometer information heavily and requires large memory space on computing devices.Therefore,it is not suitable for mapping with larger scenes;both LOAM and Cartographer are SLAM mapping frameworks suitable for outdoor environments,but LOAM has no loop detection and optimization,so the mapping accuracy is not as good as Cartographer.Therefore,this thesis' s framework is based on Cartographer graph optimization algorithm framework.the principle of the optimization algorithm to map are researched and mathematical formulas derivation in the third chapter.The thesis also uses public data in to build map with Cartographer to verify the graph optimization algorithm.3.The traditional GPS/IMU fusion framework is studied,and the algorithm is verified using public data.However,the localization accuracy is not significantly improved.Therefore,the GPS/INS/LIDAR federated extended Kalman filter based on the loosely coupled framework is studied,and the relevant iterative formulas are derived and the iterative steps are given.4.According to the GPS/INS/LIDAR integrated localization requirements,a mobile robot hardware platform were built,and the function of obtaining and processing data were verified.Finaly the GPS/INS/LIDAR federated extended Kalman filter integrated localization system based on the loosely coupled framework was implemented.This thesis contains 36 pictures,9 tables and 71 references.