Error modeling and extrinsic–intrinsic calibration for LiDAR-IMU system based on cone-cylinder features

Abstract

The extrinsic and intrinsic calibration of light detection and ranging (LiDAR) and inertial measurement unit (IMU) system is an essential prerequisite for its using in robots navigation or localization tasks. However, the existing LiDAR-IMU calibration method usually based on either point or planar features and existed adjustment parameter correlation limitations, which great restrict the extrinsic–intrinsic calibration flexibility and versatility. For this reason, a novel calibration technique that incorporates cone and cylinder features is proposed to overcome these drawbacks. The basic principle of our proposed method is that, first of all, we establish the transformation relationship between LiDAR and IMU coordinate frame, the LiDAR-IMU system calibration parameters and cone-cylinder geometric constrained. Secondly, the LiDAR extrinsic parameters are calibrated by estimating the pose of each scanned point datasets lies on the cone-cylinder surface and then solving the cone-cylinder geometric constrained optimization problem. Thirdly, the restricted maximum likelihood estimation (RMLE) algorithm is used to solve the optimization of IMU intrinsic parameters calibration Finally, intensive experimental studies are conducted to check the validity of our proposed method, the experimental results are presented that validate the proposed method and demonstrate the overall performances of LiDAR-IMU system obviously improved compared to plane based calibration method.