In CAM programming, specialized software is used to plan tool paths for CNC machine tools. The programmers manually specify a large number of process parameters, which the CAM software uses to calculate and graphically display a tool path. The CAM programmers now check, again manually, whether the calculated path is suitable for the desired result. The selection of parameters as well as the checking of the tool path is currently the responsibility of the CAM programmers and is a matter of experience. Often, many different parameters have to be tried out in several cycles until the correct path is achieved - especially if the programmers are not yet very experienced. This costs a lot of time.
The goal in the research project "CAMStylus - Development of an AI-supported virtual reality solution for the intuitive operation of computer-aided manufacturing systems" is to significantly improve the intuitiveness of tool path planning in the software and thus make it easier: In the future, CAM programmers will be able to create tool paths using gestures.
To this end, the project partners are developing a virtual reality application that allows CAM systems to be used more intuitively, as well as a neural network that is trained with images of hand movements so that it finds the correct parameters for optimal tool paths. Intuitive handling also makes it easier to train new and inexperienced employees, as they no longer have to learn all the details of the various parameters of the CAM software in question. Experienced CAM programmers also benefit from the CAMStylus system, as the intuitive handling also allows them to reach the desired tool path more quickly.
In the first phase of the project, the Fraunhofer IPT researchers are developing a VR environment and programming software for tracking gestures that reveal path movements. This not only makes it possible to record movements of the hand, but also with so-called "tracking pens", which the team is also developing in the project in addition to the VR environment and software.
In the second phase of the project, the research team is building a neural network that will derive the correct information for the intended tool path from the motion tracking data. Here, the project partners are relying on a machine learning approach, in which the neural network learns to solve a predetermined task based on a variety of training data and subsequent evaluation of the result. In the CAMStylus project, the task is to determine the correct parameters for a desired tool path from the programmer's motion data and use them for path planning. The planned path is then displayed graphically on the screen.
In this project phase, the task is to generate the training data and build a neural network that is capable of mastering this task using the training data. The success of the training is then validated by tests. In a final step, the individual components are combined into a complete system consisting of the software and the tracking pen. Finally, the project team will demonstrate the practical suitability of the overall system using application-oriented demonstrators.
The research project " CAMStylus – Development of an AI-supported virtual reality solution for the intuitive operation of computer-aided manufacturing systems" is funded by the German Federal Ministry of Education and Research (BMBF).
Funding code: 01IS22006D
DLR Project Management Agency – German Aerospace Center e.V.