Accurate Calibration of a Robotic Positioning System Using the Evolutionary Search

This paper presents an evolutionary search strategy based on the island model to find the accurate transformation matrix between an infrared optical tracking system and a 6-axis robot arm. We demonstrate how a parallel genetic-algorithm mechanism, also known as the migration scheme, coupled with a local simplex-downhill strategy can effectively solve this multi-dimensional difficult problem. Finally, experimental and numerical example of the proposed strategy for a system composed of a Mitsubishi Electric RV 6S robot arm and an NDI Polaris system is presented. Performance of the calibrated positioning system has a maximum positioning error of 0.6 mm within a one meter cubic working space