Calibration of a modular assembly system for personalized and adaptive human robot collaboration

Abstract

A modular assembly system is introduced for obtaining a high flexibility which consists of manual work stations and collaborative robot modules that can be individually arranged and connected, dependent on the actual process requirements and qualifications of workers. The modularization induces a fast adaption towards different product variants and enables personalized assistance for diversely skilled workers. The collaborative modules and manual work stations have to be both physically and virtually connected for the implementation of the system. The challenge for linking the modules is the calibration of the robot module as a function of the location of the manual work station. A vision based calibration method is implemented, using a reference object that has the same gripping planes as the assembly object. A system of infrared depth-finding cameras is used to record a reference object that is handled by the collaborative robot and to locate its coordinates by 3D-Image processing. By transferring the coordinates into a common system, existing program sequences, data for progress recognition and employee qualification can be adapted for the specific workspace, worker and assembly object in order to achieve maximum flexibility and personalization. A control program provides the interface between the robot and the infrared depth-finding camera and processes the functions of teach-in and gesture recognition. The presented paper includes the concepts of modularization and assistance, the implementation of the calibration method by 3D-Image processing and the evaluation of the system.