Robots as underwater garbage collectors

SeaClear system

Components of the SeaClear robotic system. (Image: The SeaClear Project)

The oceans are full of plastic waste. To counter this, scientists are currently developing a robotic system that can locate and collect plastic waste underwater. In contrast to many cleaning measures that are concentrated on the surface, these autonomously operating garbage collectors can remove the plastic waste deposited on the seabed – because a large part of the waste is deposited there.

The oceans of our planet are full of plastic waste – it is estimated that there are between 26 and 66 million tons of large and small plastic residues in the sea. Over time, some of it is crushed to form microplastic; together with larger residues, this garbage gradually sinks into the depths and is deposited on the ocean floor. For many animals in the ocean, the plastic parts are an acute threat, because they often eat them by mistake and then perish on them.

First map, then collect

But what to do How can you at least get the bigger plastic pieces out of the water? So far, the cleaning measures have mainly concentrated on coasts and the water surface, because cleaning the seabed is time-consuming, expensive and often dangerous for divers. That is why scientists from the Technical University of Munich are now working with eight European partner institutes in the EU project “SeaClear” to develop a robot system that will collect underwater garbage in the future.

The system is made up of four robot components: An autonomously driving robot boat carries out an initial scan of the sea floor and localizes larger accumulations of garbage. Then an observation robot is lowered into the water, which tracks down the garbage in the depth and provides close-up pictures of the seabed. If visibility is good, a drone from the air also ensures that further garbage is detected in the water. All information from these “garbage scouts” is then combined into a virtual map. This serves as a basis for a collecting robot, which then scans the points on the map and uses a gripper to place the garbage in a collecting basket.

Learning ability helps robots in the underwater world

What sounds simple in theory is not easy in practice for autonomously operating robots: “Developing autonomous robots for use underwater poses a very special challenge,” explains Stefan Sosnowski, Technical Director of the SeaClear project from the Technical University of Munich. Because unlike on land, there are very special conditions in the water. “As soon as a piece of garbage has been identified and located, the robot first has to move around it. In doing so, he can sometimes encounter strong currents against which he has to assert himself. Getting that right is our job in the SeaClear project. “

To do this, the team uses machine learning methods. Artificial intelligence learns when and under what conditions the robot has to move in a certain way in order to be able to pick up its piece of garbage. “Another challenge is that we don’t have the computing power we are used to on land,” says SeaClear project coordinator Sandra Hirche from the TU. “There is no connection to large data centers with high-performance computers. The algorithms that we develop must therefore be as efficient and resource-saving as possible. “

So far, the development has been successful despite these increased difficulties: First tests with a prototype were carried out in October 2021 in the sea off the Croatian city of Dubrovnik. In May 2022 there should be further tests in the Port of Hamburg. When the SeaClear system is fully operational, it should classify underwater waste with a forecast rate of 80 percent and collect 90 percent successfully. The robots would then be as effective as human divers.

Source: Technical University of Munich

Recent Articles

Related Stories