The Calibration Method Of The External Rotation Axis For Multi-Line LIDAR

As an advanced active sensing technique,LIDAR(Light Detection and Ranging)is a hot research topic in fields such as autonomous driving,remote sensing,and intelligent robotics.Compared to fixed-mounted LIDAR,a rotating LIDAR system not only provides a wider field of view and achieves panoramic scanning but also captures denser point cloud data.To overcome the deviation between the scanning center of the LIDAR and the axis of the motor caused by manufacturing processes and avoid inconsistencies between the reconstructed data and the scanned objects,it is necessary to calibrate the external rotation axis parameters of the rotating LIDAR system.Currently,the majority of external rotation axis calibration methods are designed for single-line LIDAR.Although these rotating LIDAR systems can compensate for the limitation of acquiring only 2D point cloud data with single-line LIDAR and achieving 3D point cloud data collection,the collected point cloud data is often sparse.If a multi-line LIDAR is used in a rotating LIDAR system,it can not only achieve panoramic scanning but also capture extremely dense point cloud data.Therefore,this thesis is directed against the calibration of the external rotation axis for multi-line LIDAR.Although the calibration method for single-line LIDAR can be directly used for multi-line LIDAR,due to the deviation between multiple pairs of laser devices within the multi-line LIDAR,if only one laser is used to collect data,it can generate significant errors.In order to ensure that the point cloud data collected by multiple laser beams can be used simultaneously during the calibration process,a checkerboard calibration board is selected as the calibration object,and an external rotation axis calibration method based on point cloud thickness optimization for multi-line LIDAR is proposed.This method utilizes the feature that each frame of multi-line LIDAR data can detect a checkerboard calibration board and converts the checkerboard point clouds in multiple frames of data into the same coordinate system.In theory,checkerboard point clouds in multi-frame data should overlap each other.Therefore,it is possible to iteratively optimize the thickness of the point cloud after the superposition of the checkerboard grids,thereby obtaining the external rotation axis parameters of the rotating LIDAR system.After systematic testing and analysis,the results show that the proposed method can effectively achieve the external rotation axis calibration of multi-line LIDAR.Compared with the current external rotation axis calibration method,the estimated parameter error of this method is reduced by 18.7%,with higher accuracy.At the same time,in order to study the impact of different rotation axis errors on the reconstruction results,different types of errors were systematically analyzed and simulated.In order to achieve external rotation axis calibration in any environment,a target-less method of external rotation axis calibration for multi-line LIDAR is proposed in this thesis.Due to the fact that each point in the environment will be scanned twice by a multi-line LIDAR after a complete one-week scan of the surrounding environment by a rotating LIDAR system,the collected point cloud data can be divided into two parts.Using the redundant information contained in the point cloud data,point cloud registration is performed on these two parts of the point cloud data to estimate the external rotation axis parameters of the rotating LIDAR system.Considering the large amount of data collected by multi-line LIDAR,register point cloud directly can consume a lot of time.Therefore,data is filtered based on the vertical field angle of multi-line LIDAR to minimize the amount of data used in the iterative optimization process.In order to verify the effectiveness of this method,randomly generated simulation data and collected actual data are used for experimental explanation.Experimental results show that compared with other current methods,the estimated parameter error of this method not only decreases by 13.8%but also reduces the overall parameter iterative optimization time by 34.2%.