Rule-based supervisory control-extension for automated manufacturing processes

Abstract

Before performing a surface finishing process, human operators analyze the workpiece-conditions and react accordingly, i.e. they adapt the contact-situation of the tool with respect to the surface. This first step is ignored in most suggested automation concepts. Although their performance is satisfactory for the general process thanks to adaptive position-and force-/torque-control algorithms, they are unable to address specific problematic cases as often encountered in practice because of variations in workpiece-dimensions or-positioning. In this work, a human mimicking element is developed to overcome this limitation of current control concepts and to translate human expertise to the robotic manipulator. A rule-based system is designed where human knowledge is encoded as if-then rules. This system is integrated with a previously suggested control strategy in a hierarchical manner. The developed concept is experimentally validated on a KUKA LWR 4+-robotic manipulator.