«SENS AI» – How we are harnessing artificial intelligence to foster the circular economy in Switzerland

Electronic waste is the fastest-growing waste stream in the world1. In 2022 alone, over 121,000 tonnes of electrical or electronic appliances were collected and recycled in Switzerland2. By automating the detection of electronic waste, we are increasing transparency in the flow of small household appliances and providing manufacturers and recycling companies with reliable and precise data through the use of artificial intelligence. SENS eRecycling is thus taking on a pioneering role in the digital transformation of the eRecycling sector and setting another significant milestone towards a circular economy.

Electronic waste is not only the fastest-growing waste stream worldwide, it also poses a significant challenge for collection points and recycling companies due to its heterogeneity. From toasters, vacuum cleaners and electric toothbrushes to electronic toys, cables and e-cigarettes: small electrical appliances represent a particularly diverse range of product classes. They also differ greatly in terms of age, intrinsic value and recyclability. Despite these challenges, electronic waste offers great potential for an effective transformation to a circular economy in Switzerland and worldwide.

Pasqual Zopp
Managing Director of SENS eRecycling
“Electronic waste is not only the fastest-growing waste stream worldwide, it also poses a significant challenge for collection points and recycling companies due to its heterogeneity.”

Manufacturers are also increasingly shifting their focus. They want to know more about their products at the end of their useful life so that these products, or parts of them, can be used again. It is therefore crucial to have an effective take-back system that is suitable for this purpose. Manual random samples, in which only one to two per cent of the flow of materials is examined, do not meet these requirements. Furthermore, the growing number of product categories has resulted in a significant increase in both time and cost for us and our disposal partners. This is why we are now relying on artificial intelligence (AI) to analyse the flow of materials more accurately.

In the past, one to two per cent of the flow of goods was inspected manually. This is no longer sufficient to meet future challenges.

‘SENS AI’ can detect 130 product categories

Launched as a pilot project in 2019, ‘SENS AI’ passed the minimum viable product (MVP) status last year. Today, the AI model has been developed to a level where it can reliably and efficiently detect over 130 product categories on the installed conveyor belts of recycling plants equipped with high-resolution camera models.

Major benefits for recycling companies

Immark in Regensdorf, the largest recycling company for small electrical appliances in Switzerland, is an important pioneering partner for the project. All deliveries of small appliances on the existing conveyor belt are analysed on site. This has provided valuable data for the project. “The project offers us many advantages. It is now easier for us to see what is actually being delivered, determine the proportion of hazardous substances, and it enables us to better estimate the value of the electronic waste.” says Sabine Krattiger, Managing Director of Immark.

‘SENS AI’ is reliable, precise and versatile

The AI consistently assigns the detected appliances to the same category, regardless of where the device is installed. This makes the data highly reliable. ‘SENS AI’ is also extremely versatile, allowing new parameters to be added easily in order to rapidly respond to the changing needs of manufacturers or recycling companies. The electrical or electronic appliances detected by the system are now assigned an average weight, recycling quota or environmental and climate benefits. This allows for better prognoses and greater insight into electronic waste. Another advantage of the AI model is that there will be no additional costs for companies if the quantity of appliances to be analysed increases in the future. ‘SENS AI’ is also designed in such a way that the system can be rolled out to other locations with minimal effort.

SENS AI requires around 10,000 images before it can reliably detect an electrical or electronic appliance. The model is trained further with each additional image.

‘SENS AI’ forms the basis for an efficient circular economy

The life cycle of electrical or electronic appliances can now be analysed in a fully automated process thanks to ‘SENS AI’. Together with our partners, we not only achieve greater transparency in the flow of materials but can now also scale the analysis process and thus make it more efficient and cost effective. In addition, the data generated through the AI model can be combined with external factors as required. As an example, it is now possible to determine the reusability of individual appliances or materials or to draw conclusions about the impact of energy policy decisions on the recycling behaviour of the population. All these measures promote and strengthen the circular economy of electrical or electronic appliances in Switzerland and beyond.

The setup

A high-resolution industrial camera is installed above a conveyor belt to record the electrical appliances. The recorded images are transferred to the cloud environment where they are detected and classified by the "SENS AI". The classified objects are enriched with additional data from other sources and made available to selected stakeholders in a PowerBI dashboard.

By using object recognition and artificial intelligence, the analysis process can be largely automated and the level of detail and informative value greatly increased. "SENS AI" is able to differentiate between 130 product classes and also identify the brand and model for selected classes.

Schematic representation of the structure of the "SENS AI".

Contact

For further information, please contact Pasqual Zopp, Director of SENS eRecycling, +41 43 255 21 90, pasqual.zopp@sens.ch.