Abstract:

Computer Numerical Control (CNC) machining consists in automating the motion control of machine tools via computers. When programming CNC machining operations, one typically writes a program using G-code commands that define the movements (linear/circular interpolations, etc.) that the cutting tool has to execute in order to shape the initial raw part into a desired final geometry, together with some other technological parameters such as feed rate or spindle speed.

Typically, CNC programming is part of the syllabus of mechanical and manufacturing engineering studies. These courses usually focus on lathes and mills, which are the most widespread machine tools that control the cutting tool in two and three degrees of freedom, respectively. After learning the basics of the G-code programming language, students usually practice CNC programming using simulators that provide virtual representations of the machining process, allowing them to check the correctness of their programs and check for undesired collisions, etc. Then, after mastering CNC programming in simulated environments, students should interact with real CNC lathes and mills. However, training with real CNC machines is not always possible due to their high cost and safety restrictions, as these machines produce large forces and project debris and chips when shaping the raw part. Moreover, when real CNC machines are available for training at universities, these are scarce when compared to the number of students that should use them, which limits their access.

To address these limitations, in this abstract we propose a remote laboratory for training CNC machining with a real machine tool. The chosen machine tool is a Stewart parallel robot, which can control the cutting tool in six degrees of freedom (three translations and three rotations). This parallel manipulator is remotely operated by the students over the Internet, where the students type their G-code commands in a user interface that sends these commands to the actual robot placed at the facilities of our university. Then, a real time camera feed shows the actual motion of the parallel robot, allowing the student to check and validate the correctness of the commanded motions. To prevent undesired damage due to incorrectly programmed commands, the real robot does not interact with a real raw part, which would also require an automated device to replace this raw part by a new one after every machining. To solve this, the proposed remote laboratory provides a digital twin of the real robot, where this digital twin repeats the movements of the real robot and does interact with a virtual raw part, showing the student how the raw part would be shaped were it actually placed in the workspace of the real robot.

The proposed remote laboratory aims to provide a valuable platform for CNC training with the following advantages. Firstly: by choosing a Stewart parallel robot as the machine tool, students can command more sophisticated and advanced motions than what typical lathes and mills can achieve. Secondly: by replacing the interaction with a real raw part by its digital twin, there is no risk of unwanted collisions and damage if the students make mistakes when commanding motions. And thirdly: providing remote control to the robot increases safety because students do not run the risk of being harmed by the machine, as well as accessibility and flexibility, since students can practice with the robot anytime and anywhere.

Keywords:

Remote laboratory, machining, parallel robot, G-code programming, digital twin.