How AI Can Help Overcome Challenges in Recycling
Recycling is fundamental to living in a sustainable world. By prolonging the lifecycle of resources already in circulation, industry demand can be satisfied while simultaneously avoiding the depletion of natural resources that is made ‘necessary’ by poor resource management. Recycling and effective resource management has a host of benefits, from offsetting the effects of climate change, protecting biodiversity, to ensuring the earth is preserved for future generations.
COP26 has brought about new commitments and action plans for governments to battle climate change moving forward. There has never been a better time to discuss the ways in which we can increase sustainability and mitigate climate change collectively. One issue is recycling, which is not only a challenge of changing individual behaviours, but also involves inspiring and motivating industries to adopt more environmentally sustainable practices.
Businesses can do great things with recycled materials:
Recycling does not mean using substandard materials or sacrificing quality, and there are fantastic examples of what can be achieved when companies source recycled materials over their new raw counterparts. One great example is clothing line Paratonia, which paired with Net Plus to make raw clothing materials out of 100% recycled fishing nets. By using recycled fishing netting for hat brims alone, 149 metric tonnes of fishing nets have been recycled, removing masses of waste plastic from the world’s oceans. Children’s toy company Green toys create toys from 100% recycled materials such as milk jugs, keeping 55 million from entering landfill.
Why don’t more businesses do this?
Despite these great examples, many businesses still opt to source new rather than recycled materials. Several challenges arise within the recycling process, incentivising businesses to opt for new sources that are cheaper, more reliable, or just generally more convenient. This preference results in new materials being created, depleting natural resources, and having reusable materials go to landfill.
Fortunately, Artificial Intelligence (AI) may offer a solution. By adopting AI into the recycling process, the cost, availability, and reliability of sourcing recycled materials could be greatly improved: making these sources more desirable to industry and ensuring that they adopt more sustainable practices. Here are some of the biggest challenges with recycling, and the ways that AI could act as a solution:
Recycling is difficult to sort and separate
One of the biggest challenges in recycling is being able to sort materials consistently and accurately into different categories. When we recycle at home, we sort according to glass, paper, plastic, cardboard, tin etc: but this is not the case when it arrives at the treatment plant. Each of these categories have a far greater range of sub-categories which they must be divided into. This is far too complicated to do within our homes – but it needs to be done so waste can be recycled. For plastic alone, there are multiple types within our standard waste, all which require a different method of treatment to recycle!
This process of sorting often requires a large manual workforce to manually categorise and separate types of waste for recycling. This is manual, difficult, and costly, and if done incorrectly then entire batches of recycled materials can be contaminated, rendering them useless. This means a lot of recycled materials needlessly end up in landfill sites. A report by the European Commission on plastic strategy shows that, although around two-thirds of all plastic packaging is recyclable, in practice just 40% is recycled. This is due to inefficiencies not only in the collection but also in its treatment and sorting. As such, it is imperative to find more effective ways to sort recyclable waste.
How AI can help:
So, how can AI help mitigate this problem? One AI driven solution that can tackle this problem is using image classification to identify types of waste within a batch of recycling. Image classification utilises convolutional neural networks (CNN’s), a network of image-data, to draw relationships and similarities within an image to identify objects within an image or video feed. In a recycling setting, this technology could take a batch of waste and classify various sub-categories within it for sorting. It can also be used at the quality control stage and used to oversee sorted batches of materials to ensure that no errors have been made, mitigating the risk of large batches being contaminated and rendered useless. This technology can also be paired with robotics, meaning that recyclable waste is identified and sorted automatically, rather than relying on a manual workforce.
By implementing AI into recycling, the process of sorting and classifying different materials can be greatly augmented to reduce costs, increase efficiency, and produce greater output of recyclable materials. As a result, recyclable materials become more cost-effective and thus more desirable to businesses against its unsustainable counterpart. By reducing the cost to recycle material and lowering rates of contamination, AI would help ensure that recyclable materials do not end up going to landfill, keeping usable materials within a sustainable cycle.
Recycled materials have an inconsistent supply
Businesses want suppliers that are consistent, reliable, and cost-effective. If it is cheaper and easier to source new over recycled materials, then that is the choice that businesses can make. This however can have devastating impacts on sustainability, such as the pandemic oil crash that made new plastic cheaper to manufacture than recycled plastic, significantly harming the amount of plastic that is recycled into new products. New raw materials offer consistency better than recycled sources, as their sources can be depleted at a consistent rate. These sources will eventually run out, meaning that this merit is only short-term, and is not sustainable in the long run.
Recycling has a high up-front cost, so large amounts of export are needed to make recycling cost-effective, which is difficult to meet effectively if inconsistent supply means export quantities are not guaranteed. New plastic can be produced as consistently as raw material can be extracted from the ground, but producing recycled plastic requires sourcing used plastic from a wider array of sources which are less consistent than new materials. The additional steps in sourcing recyclable waste makes these materials less commercially appealing for profit-minded enterprises.
How AI can help:
To solve this, treatment plants need more effective ways of managing and planning around waste supply to ensure the supply of recycled materials is more reliable and cost-effective. AI can support this in two ways. The first involves a better understanding of incoming waste through supply forecasting by using predictive analytics capability. By taking data on historic supplies of recyclable waste and using this to train an AI model, accurate and dynamic predictions can be generated for incoming waste in the future. This can then be used to anticipate future supply and plan accordingly for the most cost-effective plan of treatment of materials into recycled product based upon the anticipated supply.
The second opportunity is using AI to optimise the process of inventory management. Manual inventory tracking is time consuming, tedious, and prone to human error. What’s more, an inconsistent supply in recycling means changing demand is incredibly difficult to keep up with manually, making the problem so much worse. AI can automate this process, offering accurate and dynamic stock levels quickly and efficiently, without the risk of manual error. AI can also ensure storage space is optimised, maximising the amount of available storage for supply and export.
Both of these solutions offer immense benefits to the recycling process. By ensuring that recovery facilities have the capacity (e.g., the inventory space and the incoming supply of waste) to treat recycled material at scale, recycled materials can be produced at a more cost-effective rate. This technology would also help recovery facilities with stock building for times of lower forecasted supply, ensuring that these materials would be consistently available for businesses as a reliable source to meet demand. By helping the treatment process work towards becoming cost-effective and reliable, these materials stand a better chance in competition against new raw materials.
Recycling can be difficult to source or collect:
The logistics required to create recycled materials is huge. Rather than coming from large sources such as forests for cardboard, or oil rigs for producing plastic, recyclable waste comes from many smaller sources such as homes or communal recycling bins. All of this needs to be collected from each source so that it can be taken to be recycled, which makes the recycling process overall very costly.
These sources can also be inconsistent, meaning that a fixed collection schedule may at times be more or less than what is needed. ‘Wasted’ trips or inability to meet demand can harm the efficiency of the overall process and the effectiveness of recycled material as a supply source, which in turn effects its profitability and adoptability. What’s more, a fixed schedule means that during periods of high waste generation, recyclable material ends up going to landfill instead of being recycled, which significantly impacts environmental sustainability.
How can AI help?
Within this operation, AI can be implemented to optimise delivery logistics, thereby mitigating many of the problems with collecting recycling. Using a vehicles geolocation, an AI tool can gather information about real-time road and traffic conditions for the vehicles planned route. From this, a predictive model can generate a dynamic GPS route that is tailored to the most effective collection of waste, delivery to treatment sites, or even the distribution of recycled materials. These schedules could also be paired with supply forecasting to tailor waste collection and delivery schedules to match the forecasted demand of recycled waste.
In using this application of AI, waste collection routes could be made quicker and more effective, reducing ‘wasted’ aspects of the journey. By saving time and fuel costs through smart and dynamic route optimisation, the cost of delivery logistics can be reduced, ensuring that the process is more cost-effective and improving the desirability of recycled materials. This dynamic adaption helps to ensure excess is collected would also mean that reusable products are not wrongly sent to landfill, ensuring that materials stay within a sustainable cycle.
Final word:
With AI working towards improving the recycling process, and a larger global focus in achieving sustainable resources and practices after COP26, it is now an imperative for enterprises to adopt sustainable practices like recycling and use methods that don’t harm both their own operations and the environment. With more companies and governments developing strategies to increase their recycling practices and achieve sustainability, we can expect more businesses to follow in pursuit by adopting artificial intelligence to improve and optimise the process.
Written by Joseph Myler and Andrew Modrowski