How robots are reshaping 'dirty, dull and dangerous' recycling jobs
Dr Antonio Espingardeiro, IEEE Institute of Electrical and Electronics Engineers member, describes how robots and AI are revolutionising the recycling sector
Recycling has become one of our most important activities to support resource continuity and environmental sustainability for the next generation. With an increasing global population, and significant waste disposal demands, we are likely to see increased incentivisation for citizens to recycle more consistently and effectively in years to come.
With this, it should come as little surprise that robots are becoming increasingly commonplace in recycling facilities worldwide and there are many further deployment projects underway. Robots are much better suited to the demands and risks present in today’s recycling facilities, and there are a number of innovative technologies being used to ensure the most efficient, accurate process.
Using robotic arms and delta robots, these machines can pick and sort different types of trash. High resolution cameras located on the top of the robot point towards the conveyor belt, taking photos of the object beneath that are then transmitted to an onboard computer. From this, image recognition algorithms identify the object’s chemical composition using features such as shape, colour and size. If there is a match – for example, with plastic bottles – the computer locates the object by its coordinates and instructs the robotic arm to pick it up and place it into another container. It is worth noting that this entire process is carried out at a speed that would be virtually impossible for a human to replicate.
Image recognition has advanced immensely in the last decade in terms of its underlying classifying features – shape, size, colour and texture. Other approaches rely on a combination of optical recognition that distinguishes between plastics, and magnetic materials that pull iron and steel products from the mix. Interestingly, researchers are now exploring other forms of object detection that rely on machine learning and grasping. This means that by analysing the grasping forces, the onboard computer can identify what type of material the robot is dealing with. However, when it comes to recycling applications, image processing is a much faster way to sort items and get the job done.
Safety and efficiency aside, there are many other benefits of robots in this type of environment, namely from a productivity, continuity and resilience perspective. Robots can sort large volumes of recyclable waste almost non-stop, because they do not warrant the same considerations as a human workforce – for example, requiring breaks or taking holidays. This is especially relevant in the current climate, as robots have no health concerns, do not need to isolate and do not need to social distance.
That said, when discussing artificial intelligence or robotics in any industry, there must be careful consideration for the human workforce and the impact that automation can have. Society has advanced at a much faster rate in the past century than at any other known point in time, and it seems to be gathering pace. Robotics will accelerate this change, changing the nature of some jobs by taking over tasks traditionally handled by humans and completing them with greater speed and efficiency.
Humans are tremendously good at common sense and pondering; computers are extremely good in presenting facts, patterns and source data. So, what you are going to see is the merging between human intelligence and machine learning. Jobs where creativity, critical thinking, emotion, social intelligence and human contact are important are not likely to be fully integrated into robotics anytime soon. These are all human traits that are difficult to generate and translate through machines. Conversely, dirty, dull and dangerous jobs like recycling and waste collection are rapidly proving to be a good fit for robotics.