Artificial intelligence in research, education and industry
The use of artificial intelligence in industry, education and research is becoming increasingly important. In order to visualize this complex topic hands-on, the model Quality Assurance with AI System from fischertechnik is ideally suited. A sustainable learning experience is created thanks to the linking of theory and practice.
Visualization of quality assurance through AI with fischertechnik.
The use of artificial intelligence in quality control brings many advantages, which are already being used in the automotive industry, for example. Processes can be shortened, error rates and costs minimized, and error assessment standardized. The fischertechnik sorting system is supplied with workpieces in three different colors. These workpieces are marked with three processing features as well as different defect patterns. The workpieces are scanned by the camera and classified using the trained AI. Depending on the color, feature and defect pattern, the workpieces are then sorted by the artificial intelligence based on their quality characteristics. The AI used is implemented with machine learning in Tensorflow, where an artificial neural network was trained with image data. The learned AI is executed on the fischertechnik TXT 4.0 controller. The flow control of the model is implemented in the programming environment ROBO Pro Coding and in Python.
Generating your own AI applications
If you want to go one step further, you have the option of generating your own AI applications. The training is done in Python, for which a corresponding example project is provided for explanation.
Model structure of the sorting line with AI
Sorting line for workpieces in 3 different colors (white, red, blue), with 3 different machining features (bore, milled out, bore+milled out) as well as different error patterns (bore out-of-round, bore missing, milled out missing completely or partially, cracks in the workpiece. These machining and defect features are simulated with corresponding adhesive labels on the workpieces. The workpieces are scanned by the camera and classified using the trained AI. Depending on the color, feature and defect pattern, the workpieces are then sorted into 4 different bays. The AI is implemented with Tensorflow and is run on the TXT 4.0 controller. Own AI models can also be generated. The training is done on a computer in Python. A corresponding example project is provided. The sequence control for the sorting system is implemented in the programming environment ROBO Pro Coding and in Python.
Components
• TXT 4.0 Controller
• USB camera
• encoder motor
• compressor
• 4x 3/2-way solenoid valves
• 4x pneumatic cylinder
• 5x light barriers (5x phototransistor + 5x light barrier LED)
• 4x LED for illumination of the camera field
• 24x workpieces
