SINGAPORE, April 29, 2022 /PRNewswire/ -- Open Ocean Engineering has developed ClearbotNeo – a sleek AI-enabled robotic boat that autonomously collects tons of floating garbage that otherwise would wash into the Pacific from the territory's busy harbor. 

At just three meters long and pushed along by a solar battery-powered electric motor, the Clearbot Neo systematically skims the surface and scoops up floating trash, using AI to recognize and log the types of trash it collects and where. With Microsoft Azure, it can collect data for companies, governments, non-government organizations and individual, to track the origins of marine waste, then tackle pollution at the source.

Additionally, it can bring in as much as a metric ton of refuse per day for recycling or disposal. And when fitted with a bespoke boom, it can tackle localized oil and fuel spills by collecting up to 15 liters of pollutant a day. 

Millions of tons of plastic trash float down polluted urban rivers and industrial waterways and into the world's oceans every year. Now a Hong Kong-based startup has come up with a solution to help stem these devastating flows of waste.

After a long developmental phase, its creators are planning to scale up and have fleets of Clearbot Neos cleaning up and protecting waters around the globe.

The United Nations estimates that as much as 95% of plastic pollution in the world's seas gets there via 10 major rivers, eight of which are in Asia.

And there are fears that the volume of plastic trash flowing into marine environments could nearly triple by 2040, adding 23 to 37 million metric tons into the oceans per year. That would be equivalent to about 50 kgs of plastic garbage per meter of coastline worldwide.

"If we clean up our rivers and harbors, we are helping to clean up our oceans," says Clearbot Neo's co-creator Sidhant Gupta.

At just three meters long and pushed along by a solar battery-powered electric motor, the Clearbot Neo systematically moves up and down designated sections of water – much like how a household robot cleaner moves across a living room floor.

Unlike other and much larger marine trash collection solutions that are tackling pollution on the high seas, the compact nature of the Clearbot Neo makes it ideal for harbor, canal and river use.

It skims the surface and scoops up floating trash onto an on-board conveyer belt fitted near its bow between its dual hulls and into a holding bin near its stern.

Clearbot Neo uses AI to recognize and log the types of trash it collects and where.

It can bring in as much as a metric ton of refuse per day for recycling or disposal. And when fitted with a bespoke boom, it can tackle localized oil and fuel spills by collecting up to 15 liters of pollutant a day.

But this is more than just a simple clean-up machine. It also collects masses of data in the cloud using a two-camera detection system.

One camera surveys the water's surface so the bot can identify rubbish and avoid marine life, navigational hazards and other vessels – making it safe and versatile for river and harbor work.

The second camera photographs each piece of trash that lands on the conveyor belt and transmits its image and GPS location to the company's data compliance system, which is hosted on Microsoft's Azure platform.

When this data is put together with variables, like sea current and tide information, environmentalists and marine authorities have a head start on identifying the sources of the trash. Water quality data is also fed into the cloud.

Computer engineers Gupta and Utkarsh Goel founded their startup and began working on their Clearbot solution shortly after graduating from Hong Kong University in 2019.

Their inspiration came during a trip to the Indonesian vacation island of Bali where they witnessed how local workers would take to the water every day in small boats and even on surfboards to manually fish trash out of the sea to keep the shoreline and beaches safe and clean for tourists.

That got the two partners thinking: How could this slow and cumbersome process be automated?

Gupta and Goel developed a basic aluminum prototype in Bali and upon their return to Hong Kong, upgraded to a fiberglass version. A series of prototypes followed with the sleek Clearbot Neo being the latest model.

The most challenging part of the project was developing an AI model that could detect and identify waste in the water.

"We simply didn't have the computing power available to train, run and test the models," Gupta says. "This is exactly where Azure comes in. We ended up getting an AI for Earth grant from Microsoft in Spring 2020, and over the next year developed the AI model entirely on the Azure platform.

"It took a while because initially we didn't have enough data to reasonably train it, but very quickly we ended up building out a model. We then put it on the robot and started training it for path planning, collecting waste and generating data."

With the aid of GPS, Clearbot Neos can simultaneously clear the trash and produce a data point for each and every item collected — information that includes location, size, type, material and weight. After every mission, Azure's AI capabilities have already classified the Clearbot Neo's haul and added it to a growing database.

"We use Azure Functions, Azure Container Registry and Container Instances to help us deploy our AI models as well as run our back-end systems," Gupta says.

The hard data is actually more valuable than the physical material that is being collected. Only 20% to 40% of marine plastic waste in Hong Kong's waters can be recycled. Most of it is too contaminated or broken down to be usable.

But with Clearbot Neo and Azure, "We're finding out how the trash ends up in the water in the first place," Gupta says. "It adds a lot of transparency to the process of marine clean-up. We generate data about what's in the water, what's the make-up of the stuff that's there, how much of it is recyclable and what materials we should be focusing on."

With a wealth of information, Clearbot Neo and its target customers — companies, governments, non-government organizations and individuals—don't need to extrapolate about the scope of the problem in their waters. They can track the origins of marine waste, then tackle pollution at the source.

Recently Sino Group, a Hong Kong property company, acquired a model and will work with the Clearbot team on helping keep a yacht marina clean and with testing the technology further.

"Sino Group is committed to promoting sustainable living and green operation in the communities it operates and is keen to explore green and sustainable solutions for the property industry," says Andrew Young, who is associate director of the company's innovation department.

"We find the electric-powered Clearbot is a viable solution for the marina at Gold Coast Yacht and Country Club to automate waste collection from the water's surface with its AI and self-navigate function. It is a green solution with no noise or air pollution. So, we are pleased to collaborate with Clearbot by providing the marina as a testbed for the solution."

So far, the Clearbot Neo has been operating only in Hong Kong waters. Several potential customers in other countries have contacted Gupta and Goel, however, expansion has been temporarily hampered by pandemic travel restrictions.

Confident that they have a global solution on their hands, Gupta and Goel see a future where coordinated fleets of Clearbot Neos are constantly patrolling waterways anywhere.

"The entire challenge is that there are not enough resources to tackle the pollution problem— why is this so? Because there's no economic value there. At the moment you can add that, it's a different story." Gupta says.

"So that's where we're coming from. What I'm really happy about is we're able to do this work and do it profitably. We're able to take a broad vision and create something that's working very specifically."

Note to editors: For more information, news and perspectives from Microsoft, please visit the Microsoft News Center at http://news.microsoft.com/. Web links, telephone numbers and titles were correct at time of publication, but may have changed. For additional assistance, journalists and analysts may contact Microsoft's Rapid Response Team or other appropriate contacts listed at https://news.microsoft.com/microsoft-public-relations-contacts.