Diving robot that collects underwater trash
German researchers have developed an innovative underwater robot with artificial intelligence that could transform the way we deal with marine pollution by collecting ocean debris with its giant four-fingered robotic hand.
To tackle one of the world’s biggest environmental problems, marine litter, the engineering team at the Technical University of Munich built the robot as part of the SEACLEAR project.
The autonomous diving robot uses an integrated AI system with ultrasound and cameras to detect underwater litter. Once identified, it collects the waste and brings it to the surface.
Demonstrating its capabilities for the first time in the port of Marseille, France, the diving robot collected various debris from the seabed, including objects weighing up to 250 kg.
Targeting marine debris
With ocean debris ranging from electric scooters and lost fishing nets to old tires from harbor basins and broken glass, the system is designed to handle both heavy lifting and delicate precision.
According to Stefan Sosnowski, PhD, chair of information technology control at TUM, waste disposal using autonomous underwater collection becomes profitable at depths of 16 meters or more.
That is why the entire system consists of an unmanned service boat with a dinghy, a drone, a small underwater search robot, and TUM’s diving robot. The service boat provides power and data connections to the robots via a cable.
At the same time, it also sends ultrasonic waves into the depths to generate an approximate map of the seabed. The small 20-inch (50-centimeter) search robot then scans the seabed quickly and efficiently.
Powered by eight mini turbines, the submarine dives to the locations where trash is detected and collects the objects. It uses a claw to load them into an additional autonomous boat that serves as a floating waste container.
“Since we first have to identify the trash and picking up the objects requires a high degree of precision, we have a camera and sonar on board that allow for guidance even in murky waters,” said Sosnowski.
Unique features
Identifying underwater trash is no easy task, as there is little image data available to train neural networks. However, the team managed to label more than 7,000 images as potential objects that do not belong on the seabed.
Once the debris is detected, the AI transforms the images into 3D models. “This is important for deciding where the object can be safely grasped,” explained Sosnowski.
He said that the four-fingered hand of the autonomous claw can exert a force of 4,000 Newtons (N) and lift objects weighing up to 250 kg.
However, their grip is regulated by special sensors that measure how much force can be applied without causing damage. This prevents plastic buckets from breaking or glass bottles from shattering, for example.
Although the boat operates autonomously, researchers kept it tethered during the experiment. The onboard battery lasts only two hours, and the cable enhances AI performance and helps pull heavy objects to the surface.
Meanwhile, the 120-kg submarine is surrounded by floating foam, keeping it suspended in the water when the mini-turbines are turned off. This allows the robot to move freely and stay on course. “This is important for approaching objects with precision,” Sosnowski concluded in a press release.
