Fraunhofer Institute for Integrated Circuits IIS
Fraunhofer Institute for Integrated Circuits IIS
© Fraunhofer IIS/ Paul Pulkert

DangerSort – Sustainable and safe

© Fraunhofer IIS/ Paul Pulkert

DangerSort: Intelligent AI sorting for the early detection of lithium-ion batteries in waste streams

© Fraunhofer IIS
X-ray image of the material stream with marking of hazardous objects.

DangerSort is an integrated sorting solution installed upstream of industrial waste treatment plants which uses AI-based sorting to detect and remove hazardous objects – in particular lithium-ion batteries – at an early stage. The system addresses the increasing fire risks in sorting facilities resulting from the growing number of batteries in waste. Using a high‑throughput X‑ray system and AI‑based image analysis, DangerSort identifies batteries even when they are hidden in dense, multi‑layer material streams. By selectively removing them using pneumatic valves, safety in sorting and recycling facilities is significantly increased.

Fields of application

  • Early detection of hazardous objects (especially lithium-ion batteries) in waste streams
  • Pre-sorting upstream of industrial sorting and recycling lines
  • Increasing safety in sorting facilities by reducing fire risk
  • Testing and further development of AI detection models under real operating conditions
© Fraunhofer IIS
Mixed waste before AI-powered hazard detection with DangerSort.

Specifications

Expand all Close all

Material flow Up to approx. 30 t/h
Conveyor speed 2–3.5 m/s
Sensor principle Transmission X-ray system (XRM)
Spatial resolution Approx. 0.4 mm
Image acquisition frequency Up to 7,500 Hz
Acceleration voltage Up to 160 kV
System dimensions [L x W x H] approx. 8000 x 2700 x 2200 mm
System weight approx. 13 t
Sorting width 1600 mm; different sorting widths possible

Design and operation

© Fraunhofer IIS/ Paul Pulkert
The prototype was completed at the end of 2024. In the meantime, the system is being used in real operation.

DangerSort consists of a high‑resolution transmission X‑ray sensor, an AI‑based analysis system, and a pneumatic ejection module. The X-ray sensor continuously generates high-resolution, high-frequency images of dense, multi-layer waste streams, which are evaluated in real time by deep neural networks. The AI interprets the internal structure of the objects and identifies batteries, in particular lithium‑ion batteries, as especially critical hazards. When such an object is located, high‑speed pneumatic valves control targeted ejection, removing only a small amount of surrounding material. The architecture is designed for high throughput and robust, long-term operation and can be integrated into existing sorting lines.

At a glance

  • Reliable early detection of lithium-ion batteries in complex waste streams
  • Real-time analysis by AI models adapted to X-ray data
  • High throughput: up to approx. 30 t/h at approx. 2–3.5 m/s conveyor speed
  • Transmission X-ray with high resolution (around 0.4 mm), high image frequency (up to 7,500 Hz) and up to 160 kV
  • Penetrates dense, non-monolayer material layers and handles large, heterogeneous objects without fine shredding
  • Targeted ejection via high-speed pneumatic valves with minimal disruption of the main material flow
  • Use of industry-proven components, continuous operation and maintenance-friendly design
  • Industrial prototype completed at the end of 2024; in practical testing since spring 2025 at a Lobbe site

You may also be interested in: