Recycling in the US is collapsing and low cost robot arms like Sixi can help prevent disaster. The buyers of recyclables demand pre-sorted material with almost nothing mixed in. The suppliers are finding the cost to sort is astronomical. While much of the sorting is done by machines already, the job could be done even better with tireless robots.
How recycling works
Typically when you drop recyclable material in the blue bin it goes to a facility where it is crushed, packaged in bulk, and sold to a buyer that can repurpose the material. The largest buyer for canadian and american recyclables is China.
Recycling buyers have tightened standards
“As of Jan. 1, 2018, China, which buys approximately two-thirds of North America’s recyclables, requires that contamination levels – newspaper smeared with ketchup, plastics mixed with broken glass – can’t exceed 0.5 per cent as part of its National Sword initiative.” — The Globe and Mail
Humans sorting is too expensive
” A study by Rob Taylor with the State Recycling Program in the North Carolina Department of Environmental Quality estimated that the average market value of a ton of mixed recyclable material arriving at a recovery facility in the state dropped from just over $180 in early 2011 to less than $80 at the end of 2015. That value has since rebounded a bit, Taylor found, to a little over $100, but it still leaves the industry struggling to extract profit from the millions of tons of recyclable material Americans throw away every year. ” — USA Today
New solutions in High Tech
There are still buyers! The cost to prepare the has gone up, which lowers the return on investment for anyone trying to do business with recycling. Any technology that can lower the cost of recycling is a worthwhile purchase for these businesses. Let’s consider a typical recycling scenario and apply some new technology to the challenge. Suppose a pile of mixed recycled material being fed onto a conveyor belt.
The first problem is to identify the material so that it can be sorted. Computer vision and AI has been making major strides in recent years. Have you seen the Tesla’s object identification? Or VACnet, Valve’s anti-cheat system? Pattern recognition is something Deep Learning does really really well – in many cases, better than humans. In the early days it starts by having a Deep Learning system watch humans sort recycling while it learns what goes where. Soon enough, it knows enough that it should be able to spot things humans missed. Best of all, the Deep Learning doesn’t need to be retrained – it’s an employee that works forever and only gains experience.
The second problem is to physically separate the material that has been identified. This is where robot arms come into play. They don’t even have to be fast – if the arms are too slow, throw more arms at the problem!
Why aren’t robots recycling today?
The 2019 price of low end industrial robot arms is north of $10k USD each. Lowering the cost of robot arms will speed adoption. It is a worthwhile goal for a company that looks to the future and cares about the fate of the world.
There is no question that we have to do it. By lowering the cost of robotics we can raise the profitability of recycling and encourage less landfill. Until the world learns to stop making garbage, Marginally Clever Robots hopes to fill the gap and provide the low cost robots to help prolong the human experiment. Our customer base should be 7 billion strong.