Unconventional calibration strategies for micromanipulation work-cells

This paper presents and compares a set of calibration strategies useful to calibrate vision-based
robotised work-cells for micromanipulation and microassembly. To grasp and release microparts
precisely, robot calibration, camera calibration and robot-camera registration are needed. Conventional
calibration methods are very onerous at the microscale, therefore, two alternative unconventional
procedures, called virtual grid calibration and hybrid calibration, are developed for work-cells with
high-performance robots, minimising necessary instrumentation. Moreover, an effective calibration of
the robot end-effector is designed to compensate for misalignment and orientation errors with respect
to the vertical rotational axis. This paper describes the calibration methods and their implementation,
the results and the improvements achieved. A detailed comparison between the hybrid and the virtual
grid calibrations is provided, demonstrating the higher performance of the latter strategy.