The Research And Realization Of Indoor Localization Algorithm Of Unmanned Vehicle Based On Qr Code And Corner Landmark

In recent years,indoor unmanned vehicle technology has become a hot research.How to locate is the key point of indoor unmanned vehicle technology.We focus on researching the indoor localization technology in this paper.The indoor unmanned vehicle technology is consist of relative indoor localization and absolute indoor localization.The relative indoor localization technology depends only on encoder.While the accumulative error will increase by with the passage of time and the accuracy of localization will be lower.Absolute localization technology is different from relative localization technology whose accumulative error will be corrected by the coordinate data of artificial Landmark.WiFi,radio frequency and sonar are often used as artificial Landmark method.Among all of them,WiFi and radio frequency are based on calculating the strength of signal which are subject to interference.the accuracy of this method is also low.Although the accuracy of sonar is high,the cost of this method is more expensive than other.So we choose the localization based on qr code and corner which are flexible and easy to maintain and the accuracy of vision and raser is higher than other.The purpose of this paper is to design indoor localization system based on qr codes and corners artificial Landmark in order to solve the problem of low accuracy of indoor localization.According to the research object,we choose corresponding sensor to set up our experimental platform,including selection and equipment installation such as raser ladar,camera,IMU and encoder.then,we finish designing our experiment software system which is consist of information achievement model of sensor equipment,interaction model,localization model and control model of this vehicle.In design of localization model,we choose EKF-SLAM algorithm as our experiment system localization algorithm and reasoning and proof.Then we use camera with PnP algorithm to identify qr code coordinate information to build observation model of kalman filter.According thesis of PnP algorithm,the coordinate value of qr code must be calculated when the unmanned vehicle detect the qr code firstly so as to matching detected qr code.So we put forward a method that we combine raser lidar and camera to calculate the world coordinate data of qr code and deduce and testify.We testify EKF-SLAM system based on qr code by the data of the experiment.The experiment result show that the localization system is feasible.In order to make the indoor localization system more stable,we also take corner as landmark.We use raser ladar to detect corners.we use RANSAC algorithm to fit lines so as to improve the accuracy of linear fitting and achieved the corner landmark.Finally,after design of indoor localization system,we modified the controller of the electrical car.Then we testify our experimental localization system.The result show that the indoor localization system based on qr code and corner landmark will be more stable.The result also verify the feasibility of the indoor localization we proposed.