HomeBusinessMicrosoft teamed up with a nonprofit using autonomous 'interceptor' boats to clean up the ocean and is helping it identify trash with machine learning
Microsoft teamed up with a nonprofit using autonomous 'interceptor' boats to clean up the ocean and is helping it identify trash with machine learning
Cleaning up the oceans is a huge undertaking, especially for a single nonprofit based out of the Netherlands, but having Microsoft on your side is a nice bonus.
Boyan Slat launched The Ocean Cleanup nonprofit in 2013, with the goal of cleaning up the Great Pacific Garbage Patch. Since then, the project has also embraced the goal of preventing new waste from entering the ocean by cleaning up rivers the carry many of the pollutants.
In 2018, The Ocean Cleanup was a participant in Microsoft's annual hackathon, where volunteers work together on moonshots to try to come up with innovative solutions. The resulting machine learning models have helped The Ocean Cleanup track plastic and other waste, and informed how and where the nonprofit deploys its giant autonomous plastic collectors.
Take a look at how it works.
The Ocean Cleanup is a nonprofit taking on plastic pollution on two fronts: plastic already in the ocean, and plastic moving into the ocean through rivers.
Plastic in oceans tend to form large systems like the Great Pacific Garbage Patch and break down into microplastics, which can be harmful to marine life and eventually people.
According to The Ocean Cleanup, going after this waste with ships and nets would be expensive, time consuming, and require massive amounts of fossil fuels.
Instead, the project is working with what it calls a passive system for cleanup.
Interceptors like this one are part of the nonprofit's fleet.
The solar powered vehicle autonomously collects plastic from rivers before it can reach the ocean, moving with river currents.
With some already deployed in Jakarta and Malaysia, The Ocean Cleanup hopes to use interceptors on 1000 of the most polluting rivers around the world within five years.
Vietnam and the Dominican Republic are the next sites.
Each interceptor has a capacity of 65 cubic yards of waste and weighs nearly 50 tons.
Despite the massive size, the nonprofit says the interceptor is scalable because it was designed for mass production, can work in most of the world's polluting rivers, and requires only minimal human contact with dangerous pollutants.
The interceptor is a kind of catamaran, where water continues moving with the current while concentrated plastic flows into the interceptor.
The current moves debris onto a permeable conveyer belt, leaving only pollutants behind.
Depending on the weather, current, and other factors a single interceptor can collect more than 11,000 pounds of debris in a day.
Once full, the interceptor brings waste to shore to be sorted at a local facility.
This is where Microsoft's help comes in. In 2018, Microsoft employee Drew Wilkinson reached out to The Ocean Cleanup by email about how the tech giant might help the nonprofit at its annual hackathon.
"Microsoft has immense computational resources that could really help you track and monitor your efforts at a fraction of the cost using AI," he wrote in his outreach email.
Over two company hackathons, Microsoft employees built a machine learning model to track plastic flowing through rivers with interceptors.
Before Microsoft's help, the nonprofit had one person labeling and identifying images of debris by himself, an inefficient and tedious process.
The nonprofit shared images with Microsoft volunteers, who used the machine learning model to identify plastic versus other debris like leaves or branches.
In one summer, they were able to label more than 30,000 photos.
Now they're working on a similar model for ocean photos, too.
Vessels in the ocean have a similar cleanup method and bring back debris to Mission Control in San Francisco for assessment.
Now, The Ocean Cleanup is working to recycle the collected plastic into new products to keep the nonprofit financially sustainable.
No comments:
Post a Comment