A flexible solution to AX=XB for robot hand-eye calibration

Abstract

A direct linear closed-form solution followed by Jacobian optimization is proposed to solve AX=XB for hand-eye calibration. The approach does not require A,B satisfying rigid transformation rather than the classic ones based on quaternion algebra or screw rule, so the technique is more flexible. Rigid deduction and demonstration are given based on matrix theory. Computer simulation and real data implementation indicate that: (1)In computation of initial value, our technique has good precision and robustness. (2) As more equations are added, initial value will converge to final value gradually, which shows it credible to regard initial value as final solution when many equations are supplied.