Recovered Paper Classification System

Classification of individual paper objects

The PMV is developing a measuring system which is able to characterize recovered paper samples fully automatically with regard to their composition. For this purpose, individual objects are isolated from a total quantity, object characteristics are extracted from sensor data and the objects are assigned to a material class according to the EN 643 paper for recycling grade list with the aid of artificial intelligence.

The aim of the measuring system is to analyse recovered paper samples of up to 100 kg fully automatically with regard to their composition. For this purpose, an approach based on manual sample analysis was chosen: Each object is separated from the sample, examined and then assigned to a class. The measuring system works in the same way: The system is filled with a (partial) sample on the input side. In a first step, the paper objects are isolated and unwanted dirt particles are removed. To do this, the sample first passes through a screen drum. In this drum, small particles such as sand, glass or paper chips can be ejected from the measuring system. In addition, the sample material is clearly rectified by the drum. At the outlet of the drum, the paper falls onto a conveyor belt, at the end of which a robot transports paper objects one by one into the measuring cell. In order to be able to grip all paper objects as far as possible, a paper gripper was developed especially for this purpose. The actual measuring system contains sensors that record data on the paper objects, which in turn are analyzed by a computer system. With the help of various algorithms, characteristic features are extracted. These features are the basis for an assignment to one of the defined material classes. Trained classifiers from the field of machine learning are used for this purpose.

Demo-Video Recovered Paper Sorting System