Multi-sensor detections on input-materials with deep learning to enable informed e-waste treatment

Today, the treatment of waste electrical and electronic equipment (WEEE) is characterized by manual separation of hazardous components followed by full-product shredding and mechanical sorting in order to recover recyclates. This procedure is increasingly challenged from multiple perspectives: (1) from a worker-health and safety perspective as manual labour is required in a dangerous working environment with potential exposure to hazardous substances and more fire incidents are reported in recycling facilities due to the shredding of unidentified hazardous components and (2) from an economic and environmental perspective as reusable products or components for separate treatment are seldom identified and many critical raw materials, which are of high importance towards developing a circular economy, can be found in WEEE but are barely recovered as recyclates with today’s full-product shredding approach.

As such, the need emerges to develop product-based WEEE treatment procedures. To enable this, first information on the identity and second information on the content of a product needs to be determined in order to intelligently sort the waste stream and (semi-)automatically treat dangerous, valuable and reusable components.

Therefore, this PhD will research the applicability of state-of-the-art deep learning detection techniques to multiple image sensors such as colour and X-Ray in function of this application. In a first stage from a technical perspective and in a second stage from an economic perspective by evaluating the relationship between the estimated value of a product and treatment costs.