Conventional optical sorters that use near-infrared perform well for most products. However, some complex sorting tasks remain difficult to achieve. For example, there are still a few tricky issues when we want to remove certain contaminants that emit the same infrared signal on the spectrometers.

In this context, Pellenc ST has developed artificial intelligence modules that can be added as options to existing machines. By combining NIR/VIS and DEEP LEARNING technologies, optical sorters gain in analysis speed and can thus provide a solution for the most complex sorting tasks. In concrete terms, the neural network is trained using millions of images so that it can identify the unique characteristics of objects, such as their shape, size or other visual differences. These innovations will enable new use cases to be addressed, offering the industry new separation possibilities. Eslava was the first recycler to benefit from these brand-new AI-based sorting solutions and was able to test and validate them directly on its HDPE stream.