A Sydney Council is using artificial intelligence to reduce contamination in kerbside recycling bins.

Canterbury-Bankstown waste systems coordinator Troy Leedham will provide an overview of how the technology has revolutionised the way council is improving resource recovery at an industry conference on the Gold Coast next week.

Mr Leedham says council empties some 100,00 recycling bins each fortnight, and getting residents to put things in the right bin has been a constant challenge.

Recent audits suggest an average contamination rate of 30 per cent, which represents a considerable financial risk, Mr Leedham says.

“Any loads of recycling that are tipped at our contractor’s processing facility which are deemed to be too contaminated are rejected and then subject to significantly increased tipping costs,” he’ll tell the conference.

“It also affects our waste diversion rate and objectives of increasing our recycling ability and capacity.”

Automation

So council decided to do an audit of recycling bins and use the findings to design a community education program.

That meant sending staff out to do early morning checks of bins and log contamination in a book – a process that wasn’t just laborious and inefficient, but posed potential health risks to staff.

 Automating the detection process about 18 months ago meant staff no longer had to go out on the road.

Our education officers should be out there educating the community, not wasting half their day lifting lids and walking up and down the street.

Troy Leedham

“What we did was we installed cameras in our trucks so that when we pick up the bins the cameras take photos of what’s tipped into the truck, and then we run those photos through AI which is trained to detect things that shouldn’t be in the recycling,” Mr Leedham told Government News.

“Then we get those results and we can provide education about where the contamination is.

“We can now inspect every bin – which is about 90,000 bins every two weeks – as opposed to a couple of hundred a month.”

The data means council can tailor its education programs depending on the type of contamination occurring in specific areas, and it’s also freed up staff to do other things, Mr Leedham says.

“Our education officers should be out there educating the community, not wasting half their day lifting lids and walking up and down the street.

“Now they get that data more or less instantaneously and they can work on ways to educate residents.”

Program developed in-house

Council hopes to keep refining and improving the system, developed in-house using Microsoft’s suite of AI and automation programs, including developing better ways to store and use the considerable amount of data it produces.

There are currently over 150,000 records stored in the database, which Council plans to move to a data warehouse to prepare different uses in the future.

Other improvements earmarked for the project include boosting the accuracy of the AI and introducing additional categories of contamination.

Council is also exploring other areas where artificial intelligence can be applied, Mr Leedham said.

The idea to automate the detection of contamination arose as part of Council’s $2.1 million technology-led Closing the Loop on Waste program.