| With the development of science and technology and the continuous improvement of people’s living standard,the car has become a common transportation to travel.The phenomenon that more and more families have their own cars is making the gradual loss of physical space available in the city,different forms of road traffic are facing a swelling test.And the ability to drive safely is more demanding than ever before because of the mass use of vehicles.To resolve these issues,the research on autonomous vehicle is now becoming more and more popular.In this paper,the localization system of autonomous vehicle is studied,in order to solve the problem of localization.Only if the location information of the autonomous vehicle is determined can the function of environment sensing and road planning be of practical significance.This paper proposes an autonomous vehicle localization system based on GPS/IMU/LiDAR,the proposed system fuses the location obtained by GPS/IMU and 3D points obtained by LiDAR to achieve precise and robust localization for self-driving vehicles,the main contents of this thesis are as follows.First of all,building the map of the environment.The input to the mapping problem contains two kinds of information.The data from LiDAR provides the relative position of observed features with respect to the pose from which the data is made.The IMU provides yaw angle of the vehicle.The one-step SLAM formulation uses the data from LiDAR and yaw angle from IMU to estimate the geometric relationship between two consecutive poses.Therefore,the environmental information detected at each pose can be converted into the same coordinate system to achieve mapping.And the position of the curb line can be obtained from the mapping information.Secondly,obtain the lateral distance between autonomous vehicle and curb by extracting curb points in real time.The curb points can be obtained after processing the data of LiDAR.And then the curb points are used to fit the curve equation of the curb in the vehicle coordinate system and calculating the lateral distance between autonomous vehicle and curb.Grid map method and the geometrical features of curb are used to obtain curb points from 3D point clouds.Finally,lateral error correction.Convert the position information of the GPS/IMU fused by the Extended Kalman Filter to the global coordinate system,and calculate the lateral distance from the position to the curb extracted from map.And then process the data from LiDAR to obtain the actual distance between the autonomous vehicle and the curb.The difference of these two distances is then used to verify the accuracy of the position information and correct the lateral error of the autonomous vehicle.This paper is mainly based on Robot Operating System under Ubuntu system to achieve obtaining the position information from GPS/IMU、processing the data of LiDAR in real time、correcting the lateral error、transferring information and so on.Communication mechanism of ROS based on asynchronous data streams of Topic is used to achieve passing between messages of sensors and make sure the operation in real time and verify the feasibility of the program.The experimental results demonstrate the localization method based on GPS/IMU/LiDAR can realize obtaining the pose of autonomous vehicles. |
