On recursive calibration of cameras for robot hand-eye systems

Abstract

A preliminary study of the recursive calibration of cameras for robot hand-eye systems is presented. The approach has two advantages. First, the current estimation of the camera is available and thus helps image processing. Secondly, this approach can extend to in situ or dynamic calibration of cameras for robot hand-eye coordination tasks. The performance of the system is evaluated by simulation with data from both synthetic and real cameras. The authors discuss time-varying camera parameters and smooth input data for calibration, which issues have not previously been addressed, and a solution using adaptive algorithms is given. The framework proposed is believed to have important applications for robot hand-eye systems in unstructured environments.<>