Hand-Eye Calibration Data Sheet Template (Tsai/Park/DLT Comparison)

Hand-eye calibration establishes the spatial relationship between a robot's end-effector (hand) and a camera (eye), expressed as a rigid-body transformation matrix X satisfying AX = XB, where A is the change in camera pose and B is the corresponding change in robot base-to-end-effector pose. Tsai’s method leverages analytical solutions based on rotation axis-angle decomposition and requires at least three non-collinear poses; it excels in computational efficiency and robustness for well-conditioned setups. Park’s method uses dual-quaternion algebra to solve the AX=XB equation via singular value decomposition (SVD), offering improved numerical stability and handling of noisy or minimal pose sets. DLT, while originally developed for camera calibration, is adapted here as a non-iterative, over-determined linear solution that estimates the 4×4 homogeneous transformation directly from point correspondences—making it simple to implement but sensitive to noise and requiring careful normalization. The data sheet template operationalizes these differences by structuring inputs (robot joint configurations, image coordinates, world points), outputs (rotation vectors, translation vectors, condition numbers), diagnostics (RMS reprojection error, residual norms, covariance estimates), and metadata (calibration date, hardware IDs, lighting conditions) to facilitate ISO/IEC 17025-aligned validation and audit readiness in automated manufacturing, bin-picking, and precision assembly applications.