{"id":299175,"date":"2025-09-18T15:22:29","date_gmt":"2025-09-18T19:22:29","guid":{"rendered":"https:\/\/inspenet.com\/noticias\/german-underwater-robot-targets-the-seabed-to-combat-littering\/"},"modified":"2025-09-18T15:39:07","modified_gmt":"2025-09-18T19:39:07","slug":"german-underwater-robot-targets-the-seabed-to-combat-littering","status":"publish","type":"noticias","link":"https:\/\/inspenet.com\/en\/news\/german-underwater-robot-targets-the-seabed-to-combat-littering\/","title":{"rendered":"German underwater robot targets the seabed to combat littering"},"content":{"rendered":"\n

A team from The Technical University of Munich (TUM)<\/strong> has taken a significant leap forward in the fight against marine pollution with the development of an autonomous underwater robot<\/strong> designed to efficiently identify and collect debris from the seabed.<\/p>\n\n\n\n

The underwater robot is part of an interconnected system that includes a support vessel, an autonomous auxiliary boat, an aerial drone and a small underwater exploration robot<\/a><\/strong>. The array operates synchronously, mapping the seafloor using ultrasound and cameras<\/strong>, and then guiding the main robot to hotspots of trash accumulation<\/strong>.<\/p>\n\n\n\n

\"Autonomous
Dr. Stefan Sosnowski and researcher Tzu-Yuan Huang examine the underwater robot in the port of Marseille. Source: Technical University of Munich. <\/figcaption><\/figure>\n\n\n\n

How does the underwater robot work?<\/h2>\n\n\n\n

TUM’s diving robot, powered by eight mini-turbines, is capable of accurate maneuvering thanks to its buoyancy-neutral design, allowing it to operate with precision even in waters with limited visibility<\/strong>. Using a combination of sonar and cameras, the system generates 3D models of submerged objects<\/strong>, facilitating their identification and safe recovery. <\/p>\n\n\n\n

It is worth noting that the lack of visual databases to train AI vision systems has been a challenge, therefore, the project partners have labeled more than 7000 images to train the neural network<\/strong> in charge of distinguishing between debris<\/a><\/strong> and the natural marine environment.<\/p>\n\n\n\n

The robot has a four-fingered robotic gripper capable of exerting up to 4000 newtons of force<\/strong>, allowing it to lift objects weighing up to 250 kg.<\/strong> At the same time, it has sensors that calibrate the pressure in real time, to avoid damaging fragile materials such as glass bottles or plastic buckets. <\/p>\n\n\n\n

Wired connectivity for power and control<\/h2>\n\n\n\n

Although the system is designed to operate autonomously, the robot is kept connected by a cable to the support vessel<\/strong> to ensure a constant power supply and enable better data processing<\/strong>, especially in tasks requiring higher computational accuracy. In addition, the cable acts as a means of extraction for heavy objects to the surface. <\/p>\n\n\n\n

\n