B. Ahmed Chekh / Pablo Puerto
Tekniker e Ideko
Vision integrated system for a Robot self-calibration solution through 6 DOF uncertainty assessment
Robot controllers rely on factory kinematic models using the Modified Denavit-Hartenberg (M-DH) parameters which are not adjusted in most of the cases as far as the application does not request it. These kinematic parameters are usually defined by theoretical values that are not ensured during component manufacturing nor through the assembly process. Hence, robot absolute accuracy is not guaranteed, in the order of a 1 or 2 mm, in contrast to the achievable repeatability, in the order of 20 µm, for high-end models. With robot calibration, the robot accuracy can be improved by a factor up to 10 for stiff robot models, bringing robot accuracy closer to its repeatability. A calibrated robot has a higher absolute as well as relative positioning accuracy than an uncalibrated one which can be used for demanding robot-based manufacturing applications, mainly in tasks where static positioning of the robot is required.
The calibration process implies the following workflow: identification of robot base coordinate system, characterization of TCP and measurement of the real positions. Then, the kinematic parameter evaluation is performed by means of iterative adjustment of M-DH parameters aiming to minimize the difference between the robot positioning values and the reference values obtained by the external metrology framework. The measuring systems/sensors that are used to assist in procedures to improve the positioning of industrial robots, there is a wide variety offered by suppliers of measuring systems (HEXAGON, FARO, SA, API, etc.). In general, the most commonly used systems are portable coordinate measuring machines such as laser trackers or photogrammetric configurations. In this research photogrammetric system was studied to know the limits of this technology to solve the camera position. Usually, the vision system is positioned in front of the robot solving the relative object position with respect to the base of the robot. However, in order to reduce cost, a pattern is going to be fixed and the camera position will be the input to calibrate the kinematic parameters. The workflow was: define the uncertainty requirements for the vision system, basic concept to design the pattern, laboratory check of the concepts and virtual design of the overall process. A new method to increase the accuracy of demanding robot-based manufacturing applications is studied using a photogrammetric system.
Ahmed Chekh is Master’s in Mechanical Engineer from the University of the Basque Country. He works as a researcher at the technological center Tekniker, in the mechanical engineering unit. Realizes his development within technological bet in metrology, for the development of equipment, systems and procedures for new metrology industrial solutions, including his participation in European and Basque projects.
Dr. Pablo Puerto Manrique is a project manager at Ideko. He holds a degree in mechanical engineering (University of the Mondragon) and www.metromeet.org 2 of 5 a PhD in the high-performance manufacturing process. Since he joined IDEKO, he has been working on the development of vision measuring systems. Moreover, he is adapting the international standards of evaluating the accuracy of complex technology such as photogrammetry. He has also worked as a lecturer in the high_performance machining department at Mondragon University.