Research On The Joint Calibration Of LiDAR,IMU And Real-time Point Cloud Building Method

With the development of Internet technology and artificial intelligence technology,selfdriving cars have developed rapidly from their introduction to autonomous driving,which has become a hot topic nowadays.Driverless cars have applications in industry,agriculture,traffic safety,and environmental protection.Simultaneous Localization and Mapping(SLAM)is one of the core technologies of autonomous driving,which provides environment perception,decision making and path planning in the operation of the vehicle,and guarantees the safety of the vehicle while driving.LIDAR is an important sensor that plays a vital role in the field of autonomous driving.Compared to other sensors,LIDAR measurements are not affected by light and weather conditions,thus providing accurate distance information about the surrounding environment and guaranteeing the accuracy of LIDAR-based positioning in principle,which makes it an indispensable technological component in the field of autonomous driving.The Inertial Measurement Unit(IMU)provides real-time positional state and the IMU pre-integration provides accurate initial positional alignment for point cloud alignment and calibration of the LIDAR trajectory during operation.Therefore,LIDAR and IMU are the current hotspots for multi-sensor fusion.To address the problems of low accuracy and slow efficiency of single-sensor SLAM map building,this paper proposes a tightly coupled method of LIDAR and inertial measurement unit IMU to solve the problem of spatial and temporal synchronization of multi-sensor fusion;a new method is proposed in point cloud alignment,which can effectively reduce the time consuming point cloud alignment;in the process of state estimation in back-end optimization,iterative extended Kalman filter is used for batch In the process of state estimation in back-end optimization,iterative extended Kalman filtering is used for batch optimization and state estimation,which ensures the accuracy of the algorithm and the efficiency of state estimation at the same time.The specific findings of this paper include:(1)To address the existing single-sensor algorithm with low mapping accuracy and insufficient real-time,this paper introduces IMU on the basis of LIDAR,and carries out external parameter calibration and time synchronization between LIDAR and IMU to solve the mapping drift phenomenon generated by the unified coordinate system of LIDAR and IMU in the mapping process,completes the effect of multi-sensor fusion mapping and improves the mapping accuracy.(2)In this paper,we adopt IMU pre-integration to calculate the zero-offset error of IMU for the zero-offset error generated by IMU update.Calibrate the point cloud and provide the initial value for LIDAR odometry optimization.In the process of point cloud alignment,a new voxel i Vox point cloud alignment method is used instead of the traditional ICP point cloud alignment.It is demonstrated that this algorithm can effectively improve the efficiency of point cloud alignment without affecting the accuracy of map construction.(3)In the back-end state estimation,IEKF extended iterative Kalman filter is introduced to correct the estimated states by forward propagation of IMU data,and then motion compensation is performed by backward propagation.Then,the optimal state estimation is obtained by calculating the residuals to complete the state estimation and update of the global map,and the point cloud map data with higher accuracy and robustness are obtained.