Calibration of an intuitive machine vision system based on an external high-performance image processing unit

Abstract

the article analyzes, parameterizes, and experimentally verifies the known "Hand-Eye" approach of calibration methods between a 2D camera system and a robotic system. Thanks to our new sensing devices (in cooperation with suitable software), we can quickly and easily set up a new application. We emphasize determining, defining, and comparing their geometric quantities and field of view by machine vision at the scene. The tested methodology uses a calibration tool as a reference. It verifies the observed approach on a practical demonstration example of the assembly to recognize the learned model object for sensing (cube). Realized experiments demonstrate the presented approach’s usefulness, effectiveness, and reliability (measurement success rate 98%). The research results are based on using an external evaluation unit of the 2D camera system, its available built-in functions, and communication based on Ethernet - Non-Procedural.