Fusion-based Mapping And Research On Localization And Annotation

In the field of unmanned vehicle,high precision map has a pivotal role.In the high precision map,not only can make semantic annotation work,but also can provide a great help for the environment perception of unmanned vehicle,and also can achieve high precision positioning function with high precision map as a priori.For a long time,the construction of high precision map is a very challenging thing.In this paper,a high precision map is built to work,and in the construction of high precision map based on the realization of high precision positioning and semantic annotation.Specific work is as follows:1 Application of inertial navigation and laser radar as the main sensor,the integration of the optimization algorithm to build a high precision 3D point cloud map.With the construction of the high precision map as a priori map,a high precision positioning method based on a priori high precision map is realized.2 in-depth study of the single camera positioning method,based on the ORB-SLAM to make improvements and innovation,proposed a random fern classifier based on visual range.And the combination of vision,radar and inertial navigation information,a new method based on optimization theory is proposed.3 Introduction and application of a set of high precision map annotation system,from the map annotation,import database,real-time retrieval and so on,this system has the shortcomings of the system.And make improvements on the basis of previous.To solve the problem of searching path error problem and complex intersection is difficult to retrieve the required path of the U shaped section,and constructs a set of offline simulation system,save a lot of time for the real vehicle test,but also improve the safety factor of the experiment.In order to verify the feasibility and correctness of the improvement and innovation of this paper,this paper makes a real vehicle experiment for every technological innovation.To ensure the correctness and feasibility of the theory.