Optical identification of valuable materials on printed circuit board assemblies based on sensor fusion

Increasing waste of electrical and electronic equipment (WEEE) is a major challenge of today’s society. It affects society in several ways and needs to be solved to prevent loss of important materials and to reduce environmental contamination. Tackling this challenge requires affordable and reliable technological solutions, which enable recycling in a cheap and easy manner. One step to be taken is the recycling of printed circuit board assemblies (PCBAs), which are common in many of the high level devices such as computers or mobile phones. PCBAs include a huge amount of different components and thus belong to the most heterogeneous waste a recycler has to handle. The approach described in this paper is directed towards a reduction of the diversity by identification of specific components on the PCBA, which contain specific materials of interest. These components are then available for automated, selective disassembly. In this contribution classification results of a special descriptor developed for printed circuit boards are shown for three different classification algorithms. The descriptor is based on rather discriminative simple geometric and color features. The necessary data is obtained by a pilot setup, which is also described briefly, and processed with Waikato Environment for Knowledge Analysis (WEKA), a tool for processing big data.