Researchers at the Fraunhofer Institute for Photonic Microsystems have developed a portable technique to analyze the composition of textiles for recycling. Based on the use of an ultra-compact spectrometer connected to a smartphone, this technology combines imaging, spectroscopy and artificial intelligence algorithms.
This NIR technology not only optimizes textile recycling processes[1] portable would also enable the detection of counterfeits and could also lead to as yet unknown multiple applications.
The textile REP sector: automating sorting to make it more efficient
In France, since the creation of the REP sector for “textiles, footwear and household linen”, marketers have been subject to an obligation to treat end-of-life textile waste.
Although, in almost 15 years of existence, this REP “TLC” sector has made it possible to increase the collection rate of used textiles and develop sorting activities, the volumes collected remain insufficient: just under 35% in 2018, as the target for 2019 was set was 50%[2].
There is therefore a significant need for improvement in the collection and sorting of textiles. While new specifications have just been defined for the sector (collection rate of 60% in 2028) and the obligation has been introduced for all traders to sort textile waste from January 1, 2025[3]The industry is preparing for a sharp increase in the volume of textiles to be treated and recycled.
In order to efficiently treat these waste streams and correctly identify different textiles, increasingly sophisticated automated sorting techniques must be introduced.
A technology that combines NIR, AI and imaging for more precise analysis
According to the 2023 report on technologies for sorting, recognizing and smoothing textile materials, prepared by the design studio Terra for the French eco-organization Refashion[4]Near-infrared spectroscopy (NIR) remains one of the most interesting technologies for the automated sorting of textile waste.
The technology itself has proven itself as it has been used for several decades to sort plastic waste, i.e. materials that are chemically similar to textiles.
On the other hand, the limitations of the commonly used NIR systems are also known:
- Difficulty recognizing mixtures of more than two materials;
- Since the detection is surface-based, multi-layered textiles are recognized less well;
- Materials that are contained in small proportions or are too chemically close to each other are poorly analyzed.
Despite these limitations, NIR technology remains the most widely used and promising technology for textile sorting.
This is also the opinion of the Fraunhofer researchers, who have decades of experience in building NIR spectrometers with MEMS technology (microelectromechanical systems).
This experience allowed them to develop a miniaturized NIR system controlled by AI and linked to a database of textile materials.
This portable NIR spectrometer measures only 10mm × 10mm on one side and 6.5mm in thickness. It fits into every standard smartphone. The system uses the camera module for analysis. (Source: ©Fraunhofer IPMS)
This explains Dr. Heinrich Grüger, researcher in the Sensory Micromodules department at Fraunhofer IPMS[5]in a press release dated October 4, 2023:
“Over the years, we have succeeded in miniaturizing large laboratory spectroscopy devices using MEMS technology so that they can also be used on mobile devices.”
How does AI-powered NIR analysis work?
The first step is to take a picture of the fabric with your smartphone camera.
The AI then selects a specific point in the image data to be examined by the spectral analysis module.
The light reflected from the material is then recorded by the spectrometer module.
- Entry is through an entrance slot.
- Using a collimating mirror, it is converted into parallel beams of light.
- It is then projected onto a diffraction grating using a scanning mirror.
- Depending on the angle of incidence and exit, the grating divides the light rays into different wavelengths.
- Light reflected from the grating is directed by the scanning mirror to a detector, which detects the light as an electrical signal.
These analog signals are then converted into digital signals and then analyzed by a signal processor.
Since the spectrometer has an optical resolution of 10 nanometers, the spectrometric profile obtained allows an accurate analysis of the composition of the fibers. Using AI algorithms, it is then possible to identify mixed fabrics such as clothing made from polyester and cotton.
Multiple “portable” applications
Thanks to its small size, this miniature NIR spectrometer would be suitable for all kinds of applications related to textile identification. Equipped with such technology[6]In this way, second-hand professionals could quickly check the condition of the garment even if the label has been deleted and thus point out the relevant washing instructions to future buyers. In addition, for luxury clothing, it would also be a quick way to identify possible counterfeits.
In addition to textiles, the Fraunhofer researchers are also thinking about other possible uses, for example in the food industry or even for dermatological diagnostics under medical supervision.
[1] Near infrared
[2] source
[3] source
[4] source
[5] Institute for Photonic Microsystems
[6] Apart from economic aspects, the costs are not mentioned in the Fraunhofer press release!
For further
In documentary resources