Optimal hand-eye calibration of IMU and Camera

Hand-Eye Calibration (HEC) which is necessary in robotics is proposed to determine the relative transformations between the camera and the Inertial Measurement Unit (IMU) in this paper. We know HEC could be degenerated into the problem of solving an optimization issue for the homogeneous matrix composed of rotation and translation. After acquiring an initial estimation with Kronecker product and singular value decomposition, this paper mainly introduces the Gaussian-Newton algorithm in which the Jacobian matrix is calculated by Lie algebra to optimize the estimation. The optimization presented in this paper is executed on realistic data and the experimental results show our algorithm brings an improvement on the initial value and obtains more precise results than the other optimization using classical computation of Jacobian matrix.