The Research On Positioning Technology Of Indoor Intelligent Car

Due to the safety and reliability requirements for intelligent vehicles is very high,with a precise positioning system of intelligent vehicle can effectively reduce or even avoid the incident in the process of driving,such as vehicle rub or collision accident,so in the intelligent vehicle driving process,to its perception level,vehicles must first know oneself by various ways to the location information of the outside world.Global Positioning System(GPS)is the most commonly used Positioning technology,which is widely used in military,road engineering,automobile navigation,traffic management and other fields,but GPS has the problems of multipath reflection,large error,low frequency of signal update and so on.Therefore,people need to explore other sensors or other positioning methods to assist and enhance GPS positioning.In this paper,the positioning problem of intelligent car in autonomous navigation and other work is explored,and the positioning research work based on odometer,Inertial measurement unit(IMU)and laser sensor is carried out through the combination of theory and practice.The main research content is as follows:The kinematics model of intelligent vehicle is established to find the functional relationship between state variables and control variables.Then the sensor used in the positioning experiment was modeled and analyzed,and the main parameters of the sensor were introduced.Finally,the modeling of the whole positioning problem was discussed.The IMU calibration,linear velocity and angular velocity calibration of the smart car used in the experiment are carried out to lay a foundation for the follow-up precise positioning experiment.The forward return experiment based on the odometry was used to make a comparative analysis of the vehicle movement results before and after calibration,indicating that the accuracy of straight line movement position and rotation angle of the calibrated vehicle was improved,and the vehicle could return to the starting position more accurately,which was consistent with the initial pose.The localization problem is divided into two aspects: no map localization and map-based localization.Aiming at the positioning problem of non-priori map,odometer and inertial measurement unit(IMU)are taken as positioning sensors,and sensor data are fused according to the extended kalman filter algorithm.Through the experiment of linear trajectory and curve trajectory positioning,the error between the estimated pose and the actual pose of the vehicle is to analyze and discuss the fusion positioning effect.The experimental results show that after algorithm fusion,the vehicle can estimate its own pose more accurately,but there is a problem of cumulative error,which makes it impossible to estimate long distance pose.For the localization problem with prior map,the laser sensor is used as the external sensor to construct the map.On the basis of the completed map,the linear trajectory and curve trajectory positioning experiments are conducted by using odometer,IMU and laser sensor according to the particle filter algorithm.In order to avoid the influence of drawing error on the conclusion,the first positioning result is adopted as the optimal estimation value of the truth value,and the error between the measured pose and the optimal estimation pose is discussed and analyzed according to the commonly used positioning evaluation indexes and precision requirements in engineering applications.Experimental results show that map-based positioning can effectively correct the cumulative error of odometer and IMU,and the success rate of vehicle positioning is relatively good.