Developed in partnership with analytics specialist RTT System GmbH, this company has launched the BRT HARTNER AS Automatic Sorter, its first sensor-based and AI-supported separation platform for high-precision sorting for materials that cannot be processed using conventional separation methods. The AI can not only process all of the recorded data without loss and thus not only reliably identify the material but also differentiate objects into predefined object classes. This allows precise statements to be made about the material, ensuring that different types of plastic can be precisely differentiated for recycling purposes.
Source: Eggersmann Recycling Technology
The BRT HARTNER AS combines an acceleration belt, optical sensors and compressed-air sorting. And while many sorting systems rely on near-infrared (NIR) technologies, the AS system is equipped with the latest generation of hyperspectral imaging [HSI], says the manufacturer. This HSI camera system not only generates high-resolution images, but can also provide important information about the material properties. In addition, the use of a sensor fusion-base concept sets the BRT HARTNER AS system apart from other solutions on the market. Thanks to sensor fusion, the measured values of all installed sensors (NIR/color/3D/metal) are evaluated directly as raw data and there are no further interfaces. Thus, the data from all sensors are forwarded for analysis without loss. The detection algorithm has been trained on extensive data material and therefore proves to be extremely precise. The user can recognize the material assignment of the input material live on the system’s intuitive display. In the case of lightweight packaging, for example, not only is the material assignment shown in color, but the type of packaging is also recorded and noted. The system’s flexibility also means that new sensors can be integrated into the system at any time at the raw data level, supporting the retrofitting of 3D sensors, metal sensors or future marker technologies. If required, the standard spectral AI can also be expanded to include object AI (a color camera must be installed in this case), which recognizes and separates out precisely predefined objects. — Eggersmann GmbH, Halle, Germany