Automated assembly of products by a robot-manipulator with dynamometric control of screw joint tightening

In this work, we develop a conceptual scheme for a robotic complex to perform the installation of parts and their fastening using threaded joints by a robot manipulator. A test bench was created on the basis of a KUKA KR6 R900 industrial robot-manipulator. The robot control software was implemented using the KUKA Robot Language. The proposed actuators are the authors' original development. Two variants of the actuator with screw tightening control are proposed. The first actuating device uses the rotational motion of the robot hand flange to tighten the screws. The second actuating device is equipped, on one side, with a gripper for mounting parts on the product and, on the other side, with a drive tool with a predetermined tightening torque limit. It is shown that both actuating devices ensure a complete cycle of screwing a fastener into a workpiece, from engaging to tightening with the required torque. In the first device, the tightening force is controlled with a strain gauge, the signal from which is processed by an Arduino Mini microcontroller located in a rotating device. The signal is forwarded wirelessly to a stationary controller that sends a stop signal to the robot. It was experimentally established that, due to the presence of the wireless interface, some delay is observed between reaching the torque limit value and stopping the robot rotation during signal transmission. As a result, the value of the actually reached torque may exceed the set value by 60%. In the second device, where torque limitation is based on the current in the drive motor, the absolute error of the torque setting does not exceed 0.8 N·m in the range from 0 to 25 N·m or 3.03% (the error of standard torque wrenches is about 4%). In order to meet the requirements of state-of-the-art cyber-physical production systems, the proposed complex should be complemented with intelligent functions of controlling the process of tightening screw joints on the basis of machine learning methods.