Microsoft’s Artificial Intelligence strengthens war on plastic pollution

25 Jun 2020 --- Technology giant Microsoft and Australia’s national science agency CSIRO have formed a partnership to tackle global plastic pollution using Artificial Intelligence (AI) and smart sensor technologies. The new software will be deployed to accelerate understanding of where and how trash ends up in oceans and pioneer new methods to evaluate the full life cycle of plastics to generate targeted waste management strategies and intervention methods. A recent CSIRO survey analyzing 33.5 million pieces of trash spanning 86 countries found that seven out of the 10 items recorded were wholly or partially made of plastics, notably plastic bags, beverage bottles and food wrappers.

“CSIRO’s partnership with Microsoft will help us deliver pioneering digital solutions to reduce plastic pollution entering our environment and expedite our data processing,” Dr. Denise Hardesty, Senior Research Scientist and team leader at CSIRO Oceans and Atmosphere, tells PackagingInsights. “Being able to count and track plastic debris will enable quantifiable assessments of the effectiveness of waste management programs.”

Specifically, AI technologies will target marine debris by analyzing videos of rivers and stormwater drains to identify and track garbage flows into waterways and inform intervention efforts, like the placement of river waste traps and reverse vending machines where the public can recycle bottles and cans in return for cash.

The partnership will also contribute to CSIRO's managed data ecosystem and digital academy – projects that are transforming CSIRO's digital landscape with new technologies, data capabilities and skill sets – and bring Microsoft's latest digital technology to CSIRO's wide research portfolio.

Artificial Intelligence to the rescue
CSIRO maintains that by leveraging Microsoft Custom Vision, plastic debris in marine environments can be counted much faster and used in areas that are unsafe for manual collection and counting. Leveraging Microsoft services will also enable the creation of a new application to significantly improve the current method of data collection using hand-written datasheets.Microsoft technology could capture what waste items are present before and after container deposit schemes are introduced.

A key focus of the project is using AI and cameras for accurate object recognition. “Pollution can be challenging to track as there is so much of it and there are so many different types of trash – it comes in all shapes, sizes and colors. From tires to lightbulbs, beverage bottles to plastic bags, fishing line to lids and lolly wrappers, it’s a lot to process,” Hardesty explains.

“We aim to install cameras under bridges to look at trash flowing along rivers. This will provide key information on the types of trash, how much there is and where it travels. It will also have the capabilities to narrow in on specific details, such as the label on a drink bottle or plastic bag,” she adds.

CSIRO points to research estimating that around 40 percent of waste pollution on land originates from the beverage industry – such as water bottles and soft drink cans. Working with Microsoft through its object recognition software will enable the researchers to identify trash down to the level of detail on labels and packaging.

The project’s next steps
In addition to AI technologies for object recognition, CSIRO identifies a strong correlation between trash entering stormwater drains and trash ending up in oceans. According to federally-funded scientific researchers, 40-60 percent of trash in oceans may move through stormwater drains.

“We plan to use smart sensors in stormwater drains to develop intervention strategies before trash ends up in our marine environment. Installation of video camera traps will collect information in near real-time, tell us where to place intervention points to stop leaking, and provide data on the effectiveness of interventions,” Hardesty continues.

CSIRO indicates that 40-60 percent of trash in oceans may move through stormwater drains.“Monitoring stormwater drains for plastic pollution will support the development of targeted land waste management strategies to ultimately avoid it ending up in our oceans where it can impact marine life, and create other environmental hazards.”

Understanding human behavior
CSIRO advocates for plastic packaging reduction that maintains food security, and works with industry to develop innovative circular economy initiatives that transform plastic from waste into a commodity. Hardesty highlights that adopting new ways to reduce, reuse and recycle plastics will deliver new opportunities both for business and the environment.

“Key change is also through decision-making, so understanding human behavior and consumer choices. By collecting data on the types of trash, how much there is and where it has come from, we can work with governments and industry to generate new opportunities to improve the entire life of plastics through the supply chain. This includes targeted waste management strategies, policies and incentives for better manufacturing and development of new materials,” she tells PackagingInsights.

“Utilizing Microsoft technology has the potential to deliver wide-reaching and real impacts to support national efforts to tackle marine debris. For example, New South Wales will be implementing a container deposit scheme in 2023. This technology could provide the data to evaluate the effectiveness of the scheme.”

Hardesty elaborates that Microsoft technology could capture what waste items were there before and after the container deposit scheme is introduced. This would also allow for robust assessment and evaluation of the effectiveness of the scheme and where more targeted waste management strategies might be needed.

“It’s as clear as saying, ‘This is how much trash you've saved going out to the sea through this simple intervention point and here's a pinch point where we can make a real difference,’” Hardesty concludes.

By Joshua Poole

To contact our editorial team please email us at editorial@cnsmedia.com