»BREATHE«: Fire prevention in recycling plants through sensor-assisted removal of lithium-ion batteries

According to a study by the EuRIC association, 53% of the recycling plant operators surveyed experience small fires on a daily to weekly basis. These pose a risk to both recycling plant staff and emergency personnel through burns and toxic fumes, and cause critical recycling infrastructure to fail. Residents have to be evacuated due to the risk of the fire spreading in order to avoid endangering the surrounding civilian population. These fires are often caused by incorrectly disposed old electrical appliances in conventional household waste, which are damaged during the recycling process and catch fire.


The approach of the »BREATHE« project is to remove fire-causing hazardous substances from the manufacturing process as early as possible. The initial focus here will be on sorting plants for lightweight packaging. However, the solution is designed in such a way that it can be used for other sorting plants (residual waste, commercial waste, paper, glass, etc.).


The research consortium consists of the Würzburg start-up WeSort.AI GmbH, which will focus on generating and annotating the data set and developing the AI algorithms, and the Fraunhofer Development Center for X-ray Technology EZRT, which will be responsible for the multi-energy X-ray sensor technology, the corresponding imaging processes, the mechanical structure of the demonstrator and the mechanism for discharging.

© Grispb - stock.adobe.com
Fire department in action

Project goals

© malp - 123rf.com
Li-ion batteries

The aim of the project is the automatic process-integrated detection and removal of lithium-ion batteries (LIB). For this reason, a demonstrator is being developed to show the effectiveness of such a solution in an operationally relevant environment. In a two-stage process, the old electrical appliances with LIBs are sorted out of the process.

The combination of multi-energy X-ray images (ME-XRT) and innovative deep learning methods (artificial intelligence) can be used to detect LIBs. The second step involves the development of a separation process that is capable of sorting a wide variety of waste electrical and electronic equipment from a diverse and complex material flow without generating too much additional waste.

Project details

Project: Fire prevention in recycling plants through sensor-based removal of lithium-ion batteries using multi-energy X-ray images and novel deep learning methods.

Funding: Federal Ministry of Education and Research, funding guideline »KMU-innovativ: Research for civil security«

Duration: 1.3.2023 - 28.2.2025

This might also interest you:


Sensor-based sorting

Dual-energy processes are used to identify and sort raw materials, especially in the recycling sector.