Research On Marker Calibration Method Of Location Map Based On Combination Constraints Of Line And Plane

With the rapid development of science and technology,autonomous vehicle has gradually become one of the most important research directions in the field of automotive engineering.Automatic driving vision positioning technology is an important part of intelligent vehicle environment perception,and it is also the basis for vehicle decision planning and motion control.Compared with the traditional visual SLAM,label positioning has better performance in running speed,positioning accuracy and robustness.The positioning accuracy of the tag is highly dependent on the mapping accuracy of the tag.Due to the influence of sensor noise and other factors in the mapping process,the position and orientation of the tag in the positioning map will inevitably have errors.Therefore,in order to ensure the high accuracy of label location,it is usually necessary to calibrate the labels in the location map.In order to solve the above problems,this paper proposes a label calibration method based on the line plane combination constraint.This method constrains both the side length and the plane of the label,so as to effectively improve the accuracy of the position and orientation of the label.The main research contents of this paper are as follows:Based on the improvement of the current mainstream tags,the combined tags of CVM(Combined Various Markers)are designed,and the mapping and localization methods of this type of tags are established.Perform 3D reconstruction of tags randomly distributed in space,and by specifying a unique reference coordinate system,obtain the 3D coordinates of each corner point of all other tags in the reference coordinate system.Through experimental verification,using the reconstructed label map,when the camera detects any label,the current absolute pose in the reference coordinate system can be obtained.A plane parameterization method based on vertical points is designed,which effectively solves the rank dissatisfaction problem of the Hessian matrix caused by overparameterization,and has a clear geometric interpretation.A positioning map label calibration method based on line-plane combination constraints is constructed.First,a brief introduction to the optimization method in SLAM is made,and then the performance of the three mainstream nonlinear optimization solution libraries g2 o,GTSAM and Ceres under different data sets is evaluated through experiments.Finally,line constraints,surface constraints and line constraints are designed based on the experimental results.Face Composition Constraint Algorithm.Relying on the existing conditions of the research group,an experimental platform was built,and the vehicle camera coordinate system was defined as the vehicle body coordinate system according to the research characteristics of the research project.The performance of the algorithm is tested and verified in three different scenarios.The experimental results show that the algorithm in this paper can significantly optimize the label pose error,so that the label meets the vehicle positioning accuracy requirements.