The improper disposal of electronic waste is becoming a growing issue, with many devices being discarded alongside regular plastic waste. This is particularly hazardous when these electronics contain lithium-ion batteries. If damaged during the recycling process, these batteries can ignite and cause fires in sorting plants.

In Germany, over 10,000 fires occur each year in waste sorting facilities, and approximately 80% of these incidents are linked to lithium-ion batteries. These batteries are commonly found in items such as smartphones, electric toothbrushes, and musical greeting cards. When thrown away with packaging waste, they can be crushed or punctured during sorting, leading to fires that cause an estimated one billion euros in damage annually.

To address this problem, researchers at the Fraunhofer Institute for Integrated Circuits IIS have launched the DangerSort project. This initiative focuses on developing a sensor-based sorting system that uses X-ray technology and artificial intelligence to detect and remove hazardous lithium-ion batteries from the waste stream before they can cause harm.

The system is designed to prevent fires, offering a proactive solution rather than relying solely on fire suppression systems. It also supports more effective battery recycling, contributing to a closed-loop product lifecycle.

The DangerSort system uses a high-speed conveyor belt that transports waste at speeds of up to three meters per second. Positioned above the belt is an X-ray source, similar to those used in airport security, which scans the waste as it moves. An X-ray detector located underneath the belt captures continuous radiographic images of the materials passing through.

These images are analyzed in real time using an AI system adapted from autonomous driving technology. This AI has been retrained to detect electronic devices containing lithium-ion batteries, even if they are hidden within other waste.

Based on the AI’s analysis, compressed air nozzles are activated to eject the hazardous items from the conveyor belt and direct them into a separate chamber. These nozzles, which are only about five millimeters wide, must be precisely timed to accurately remove items of varying shapes and sizes—from small button batteries to large e-bike power cells.

One of the system’s key challenges is accurately identifying and isolating batteries of different sizes and configurations. Despite these challenges, the technology shows significant promise in improving both safety and efficiency in waste recycling.

Currently, the sorting system is undergoing tests at Fraunhofer IIS. It is scheduled for delivery to waste management company LOBBE in early June 2025 for practical testing. The DangerSort project will continue through the end of August 2025.

This innovation is part of the AI Application Hub on Plastic Packaging, a broader initiative involving 51 partners from industry, research institutions, and civil society. Through collaboration in the KIOptiPack and K3I-Cycling labs, the project aims to promote the use of AI for a more sustainable, circular economy focused on plastics and electronic waste.

By Impact Lab