Spatiotemporal Calibration of Camera-LiDAR Using Nonlinear Angular Constraints on Multiplanar Target

The combinations of data captured from diverse sensors can yield reliable geometric and semantic information. However, calibration process must be preceded to make use of combined data. Calibration methods are classified by the adjusted parameters. Spatial calibration, which merges data to a single spatial reference frame, adjusts spatial parameters. Moreover, temporal offset parameter must be adjusted for sensors to acquire data under dynamic conditions, which is known as temporal calibration. In this study, a novel constraint adjustment method was proposed for spatiotemporal calibration, which combines mobile light detection and ranging (LiDAR) sensor with visual camera. It estimates seven parameters-three lever arm parameters, three bore-sight parameters, and temporal offset parameter-simultaneously. A nearly orthogonal multiplanar chessboard target was built for nonlinear angular constraints. B-spline interpolation method was employed to interpolate the plane parameters obtained from the camera and facilitate estimation of the temporal offset. Additionally, the calibration results were evaluated using correlation values and standard errors of the estimated parameters. The sequential quadratic programming method, which yields robust results for nonlinear optimization problem, was used to optimize a point-to-plane distance model subjected to nonlinear angular constraints. Assorted experiments with different sensor motions were conducted to evaluate the effects of the nonlinear angular constraints. Compared to the results of nonconstrained calibration methods, the results of the proposed method indicate lower correlation values between the calibration parameters, reduced standard errors, and fewer iterations in the optimization process, which validates the superiority of the proposed calibration method.