Recycling robots: Using AI to solve one of the world’s biggest problems

World’s recycling problem is reaching the tipping point

Humans are bad at recycling.  Even those of us who try our best often get confused when our trash is made of multiple components, and much of the recycling that ends up in recycling plants can’t be recycled because it’s too dirty with food or liquids.  This means that once our “recycling” gets shipped to a recycling plant, human workers have to sort through all of it again and pick out the items that have been incorrectly sorted.  Not surprisingly, waste management is a field with a very high drop-out rate of human workers due to high injury and fatality rates, not to mention the incredible mind-numbing drudgery of sorting recycling by hand.

Despite all our efforts, only 14% of the world’s plastic waste ends up getting recycled.  In response to the growing recycling crisis and China’s recent refusal to accept any more plastic waste, countries around the world have been scrambling to find a better way to sort their garbage.  Apart from the obvious choices like incinerating plastic waste or sending it to other developing countries like India and Pakistan, some start-ups are looking to AI in hopes it can help with the recycling problem.

Behind the scenes, AI is already being used to sort garbage

Deep-learning robots are gradually being implemented in recycling plants.  A robot developed by AMP Robotics learns from thousands of examples of a person sorting waste in a recycling plant until it’s able to sort products by itself.  The AI’s recognition of waste products is based on clues like logos, shapes and textures, which eventually learn to identify objects correctly even if they’re twisted, dirty or otherwise deformed.

It’s a similar story with Max AI, developed by the UK-based Green Recycling.  Using multi-layered neural networks and a vision system, Max learns to identify objects as it’s processing them.  The system is already being used successfully in Essex.  In both cases, the robots are able to pick out objects with a robotic arm, a vast improvement on the clumsy mechanical sorters that preceded them.

Helsinki-based ZenRobotics also has a line of recycling-sorting robots.  Developed in 2011, the robots use computer vision, machine learning, 3D laser cameras and spectroscopic cameras to sort recycling on a conveyor belt.  More recently, Barcelona-based Sadako Technologies has brought innovations to the field, including cloud technology.

Vancouver-based start-up takes recycling bots to the front lines

Canada produces more garbage than any other developed country, so instead of waiting for inaccurately sorted waste to arrive at the recycling plant, Hassan Murad and Vivek Vyashave developed a bot to help people decide which bin to put their recycling in.  Their start-up, called Intuitive AI, uses an AI sensor called Oscar (after the Sesame Street character Oscar the Grouch, who lives in a trash can) which attaches to any bin.  The camera and ultrasonic sensor can detect when someone is approaching the bin, zoom in on the item in their hand and predict which section of the bin to put the item in.

The sensors are so fine-tuned that they can even tell if the product has different components, like a coffee cup with a stir stick, paper sleeve and lid, and sort these items accurately.  Likewise, even if the product doesn’t resemble its usual self – say, a rotten banana – the AI will still recognize the product and direct the user to dispose of it in the appropriate bin.  Murad and Vyashave expect their invention will save thousands of dollars in waste management in places like fast food joints, cities, airports, malls and universities.

In a similar vein, Polish startup Bin-E has developed a bin that automatically sorts trash as it is fed in.  As a user, all you have to do is throw in a piece of garbage and wait for the AI to recognize it.  Once you confirm that the object has been correctly identified, Bin-E will put it in the appropriate compartment.

AI has an important role to play in recycling

Ultimately, AI-infused recycling robots will lead to increased efficiency in recycling plants, while sparing human workers from a dangerous and tedious job.  Most important will be the effect on our planet if we manage to sort out our ineptitude with recycling.  It’s a beautiful example of ethical AI at work.

So, are robots taking our jobs?  Yes, but only the ones we didn’t really want in the first place!