Fraunhofer Researchers Develop AI-Based Camera for Real-Time Plastic Analysis
Researchers at the Fraunhofer Institute for Photonic Microsystems (IPMS) have unveiled a compact, AI-powered hyperspectral camera designed to revolutionize the sorting of polymer materials. Addressing a critical bottleneck in the circular economy, this new optical system enables the real-time identification of plastics without the need for an external computer, streamlining quality control and recycling operations.
Precision Through Spectral Analysis
The newly developed system captures data within the near-infrared (NIR) spectral range, specifically between 950 and 1700 nanometers. Unlike conventional cameras that rely on visible light, this hyperspectral device detects the unique chemical “fingerprints” of materials. By analyzing the absorption behavior of molecules, the camera can accurately distinguish between visually identical polymers.
This capability is particularly vital for the bioplastics sector. Distinguishing biodegradable materials, such as Polylactic Acid (PLA), from conventional plastics like Polyethylene Terephthalate (PET) has historically been a challenge for standard sorting infrastructure. The Fraunhofer IPMS solution ensures that bio-based materials are correctly separated from fossil-based counterparts, preventing contamination in recycling streams and ensuring higher purity recyclates.
AI at the Edge
A defining feature of this innovation is its integration of “Edge AI.” The system processes spectral data directly on the device rather than transmitting it to a cloud or remote server. This localized processing significantly reduces latency, allowing for instant decision-making on fast-moving conveyor belts or within handheld mobile analyzers.
According to the research team, the compact design allows for easy integration into existing industrial setups. By combining a tuneable micro-electro-mechanical system (MEMS) scanner with advanced machine learning algorithms, the camera offers a cost-effective and energy-efficient alternative to bulky laboratory spectrometers. This development marks a significant step forward in automating the lifecycle management of complex polymer mixtures.
Source: Read more