Waste recycling

It’s a messy job, but AI can help

Sadako Technologies teaches BHS robots to sort recyclables

Get busy
Max-AI© powered by Sadako Technologies. Courtesy of Sadako Technologies
Max-AI© powered by Sadako Technologies. Courtesy of Sadako Technologies

Artificial Intelligence (AI) is one of the “big things” of our era, and at least two points clearly emerge from the vibrant debate of the last years: it is here to stay, and it is a powerful tool that we will be, and we are, using to achieve great results in an amazing number of use cases. Can we already take advantage of AI to make our future more sustainable? Sadako Technologies, teaming with BHS – Bulk Handling Systems, proved that it is possible and, besides, profitable.

The Spanish company is providing BHS robots working in waste-processing plants with the ability to sort garbage on moving conveyor belts. Such robots become able to distinguish a number of recyclables, in particular plastic items like detergent containers, bottles and similar, so that they can be picked and separated from the rest of the waste before they get destroyed. This process is intended to optimize waste management in the plants where it is applied, and to generate revenues for the plants themselves.

Recyclables are valuable goods in the market indeed. So every time a plant succeeds in sorting one item out of the general waste and make it available for recycling, it is generating money. The news about Instanbul having installed vending machines that offer metro credits for recycled bottles and cans, which appeared in the media worldwide last year, shows how every piece of them counts.

Computer vision

Sadako Technologies developed a computer-vision system that is grounded on AI or, more specifically, deep-learning algorithms. Their basic assumption is that teaching a machine a complex task like identifying specific items in a moving stream of objects cannot be achieved with old-style rule-based algorithmic implementations. Deep learning (which is formally a subset of Machine Learning, or ML)  came to the rescue, as it allows machines to learn by themselves what to do. In such a ‘supervised learning‘ framework, the algorithms that provide robots with the skills they need to detect the presence of, e.g., a plastic bottle among all other pieces of waste, gets fined-tuned by experience.

The company claims that its ‘algorithms and real-time massively parallel software are able to detect and recognize very difficult objects in very complex environments’, to the extent that they are outperforming human performance in some cases! Sadako Technologies makes use of state-of-the-art multi-layer neural-network algorithms (that is what Deep Learning corresponds to), trained on ‘many terabytes of proprietary labeled and segmented images’ and running on cloud-based GPU-accelerated parallel high-performance computing.

Computer vision achieved by Sadako Technologies’ software. Courtesy of Sadako Technologies


Max-AI© powered by Sadako Technologies. Courtesy of Sadako Technologies
Max-AI© powered by Sadako Technologies. Courtesy of Sadako Technologies

As pointed out above, Sadako Technologies’ AI system powers Max-AI© Autonomous Quality Control system, a machine developed by BHS, so it provides the robot with eyes and brain. The partnership between the Spanish and the US-based company made possible the first unit of Max-AI© Autonomous QC to start running in an MRF in Sun Valley (California) in 2017. Since then, many big players in the waste industry have been adopting it, making operations in plants safer and more profitable.

Sadako Technologies is also receiving funding from the European Union to develop a waste-monitoring system grounded on computer vision, with the aim of optimizing sorting tasks in waste-processing plants.  This project is called RUBSEE, and it is financed under SME-Horizon 2020 program.

How to support

  • You can follow Sadako Technologies on
  • You can support Sadako Technologies even by simply sharing this article on your favorite social networks. Spread the word!

Leave a Reply

Your email address will not be published. Required fields are marked *

I accept the Privacy Policy