Advancing Accessible Underwater Robotics: The Mini-Girona I-AUV at RAMI 2025

The ocean remains one of the last great frontiers for human exploration, yet sending machines to work there is notoriously difficult and expensive. For decades, the robots capable of performing complex tasks underwater—such as fixing broken pipes or inspecting delicate structures—have been either simple remote-controlled devices that require a human to steer them constantly, or massive, autonomous machines that cost hundreds of thousands of dollars and are out of reach for most universities and researchers. This financial barrier has kept advanced underwater robotics in the hands of a few wealthy institutions, slowing down the pace of discovery and innovation. The field is now looking for a middle ground: a machine that is affordable enough for widespread use but smart enough to operate on its own, navigating the murky depths without a human hand on a joystick.

A team of researchers has taken a significant step toward this goal with the creation of the Mini-Girona, an autonomous underwater vehicle designed to be both accessible and capable. Unlike the traditional, torpedo-shaped robots built for speed and long-distance travel, this new machine prioritizes agility and the ability to hover in place, allowing it to perform delicate work close to underwater structures. The team built the robot with a focus on keeping costs down, aiming for a price tag of approximately $50,000, a fraction of the cost of comparable research vehicles. To achieve this, they integrated advanced components like a robotic arm for grabbing objects, stereo cameras that act like human eyes to see in three dimensions, and powerful computers capable of running artificial intelligence to help the robot make decisions on its own. The goal was not just to build a cheaper robot, but to create a tool that could bridge the gap between basic remote controls and prohibitively expensive research platforms, opening the door for more students and scientists to experiment with underwater automation.

The true test of this design came when the team entered the Mini-Girona into the RAMI 2025 competition, a high-stakes event held in the marina of La Spezia, Italy. The competition simulated a real-world emergency where critical underwater infrastructure had been damaged, and teams were tasked with deploying robots to assess the situation, locate a broken pipe, and perform a repair. The challenges were severe: the water was murky with visibility ranging from just one to three meters, and the robots had to navigate a 50 by 25-meter area without human intervention. The tasks required the robots to find colored buoys, follow a pipeline, identify damage markers, and physically turn a valve to stop a leak. It was a rigorous environment designed to push the limits of what autonomous machines could do in the real world, far away from the safety of a laboratory.

The Mini-Girona faced significant hurdles right from the start. On the first day of the competition, high temperatures caused the robot's internal electronics to overheat, disrupting its ability to communicate and navigate autonomously for the first two days. Furthermore, strict security rules at the venue meant that only three of the eight team members were allowed to enter the competition area, forcing a small group to manage complex operations under immense pressure. Despite these setbacks, the robot's underlying design proved resilient. The team managed to get the robot to function well enough to secure second place overall in the competition. They excelled specifically in tasks requiring vision and physical interaction, earning top marks for their performance in inspecting the site and executing the intervention.

The robot's success relied heavily on how it "saw" and understood the underwater world. Instead of relying on a single camera, the Mini-Girona used two cameras mounted side-by-side to create a three-dimensional view of its surroundings, much like human binocular vision. This system allowed the robot to build a detailed map of the seabed, identifying objects like pipes, valves, and buoys even in poor visibility. The robot's software used artificial intelligence to recognize these objects and calculate their distance, guiding the vehicle to move safely around obstacles. When it came time to perform the repair, the robot used a five-jointed robotic arm to reach out and turn a valve. While the final turn was completed with some human assistance due to the earlier technical issues, the robot successfully grasped a retrieval pole and surfaced, demonstrating that the machine could physically interact with its environment with precision.

The results of the competition highlighted both the potential and the current limits of affordable underwater robotics. The Mini-Girona proved that a low-cost platform could compete with much more expensive machines in complex, real-world scenarios, achieving a high level of autonomy in navigation and object detection. The team's performance in the vision-based tasks was particularly strong, showing that advanced computer vision can work effectively even in the challenging conditions of the open sea. However, the experience also revealed areas for improvement. The overheating issues and the difficulties faced by the small team underscore the need for better thermal management and more robust contingency planning for future missions.

Looking ahead, the researchers plan to refine the robot's capabilities to make it even more reliable. Future work will focus on improving the robot's ability to reconstruct 3D images of the underwater environment using deep learning, which could lead to fully autonomous intervention tasks without any human help. The team also intends to test different combinations of sensors, such as pairing a single camera with a sonar system, to develop even more advanced ways of fusing data. By continuing to push the boundaries of what is possible with affordable technology, the Mini-Girona project is helping to democratize underwater robotics, ensuring that the tools needed to explore and protect our oceans are available to a much wider community of scientists and engineers.