Artificial vision method for automatic robot code generation in robotic manufacturing for automotive industry

Manufacturing operations performed by robotic systems in Mechanical Industry needprecision to achieve an high final quality on workpieces which are, generally,characterized by complex geometrical profiles, variable within small batch lots. Eachrobot movement is composed of hundreds of spatial points, calculated with respect toconstrains imposed by tool dimension and envelope. Consequently robot programmingrepresents a time consuming task, to be realized interrupting the system production. Anovel approach to the problem is discussed in the present paper. Artificial visiontechniques are applied to on-line reconstruct the workpiece profile and to calculate atailored tool path. A high resolution vision system moved by a preprogrammed roboticarm catches the workpiece. Images are elaborated and assembled to obtain a single highresolution view of the workpiece and to match the manufacturing features. Tool pathsare finally defined coherently with tool characteristics. The robot code is automaticallygenerated and transmit to the robot controller, ready to be executed. For an effectiveimplementation in Industry of the method developed, a graphic-based softwareapplication was realized. A tailored graphical interface was designed to guide the finaluser along the process and guarantee the best usability. State of the art hardware, visionlibraries and last generate anthropomorphic industrial robots were adopted to test themethod while an high performance engine head was chosen as benchmark forexperimental validation