How researchers are helping robots think for themselves to protect the environment

When a small battalion of 100-pound robots appeared in the lake behind FIU's Green Library, onlookers had no idea they were witnessing the future of marine environmental monitoring. To casual observers, it seemed like autonomous vessels moving at random and nearly colliding with each other.

But Paulo Padrão knew what was really happening.

The recent computer science graduate student was testing how well robots programmed with AI could scout a body of water without continuously communicating. The near collisions weren't accidental. Rather, the robots’ algorithms were directing them to intentionally sweep close to each other to scan as much of the water as possible.

"Buoys and other traditional water monitors only tell you what's happening in one spot," Padrão explained. "But when we put AI-powered robots in a lake, they can map the entire area."

But that’s only the surface of what today’s advanced technologies are enabling in the robotics industry, says Padrão, who graduated with a Ph.D. from the Knight Foundation School of Computing and Information Sciences in April. He sat down with FIU News for a Q&A to discuss his research.

What excites you about marine robotics today?

What's really exciting is how AI breakthroughs are making our systems truly autonomous.

Instead of pre-programming every movement, our robots can now learn and adapt, making decisions based on the data they're collecting. We literally just press start and they navigate to different locations based on what they discover.

Another game-changer is digital twin technology, which allows us to create high-fidelity simulations that researchers worldwide can access remotely.

What is a digital twin?

A digital twin is like a video game version of a real-world system. It allows us to run experiments of a system on a computer before testing it in real life.

Having a digital twin of a marine robot is incredibly useful. Think about the logistics of this research—the personnel, the weather conditions, the logistics. These vehicles are heavy.

Can you imagine moving them daily, putting them in a lake, collecting data and then hauling them back out? Our digital twin facility saves us a lot of this time, money and effort, allowing us to research more efficiently.

We created a digital twin of one of our facilities? How?

Yes, with support from the Department of Defense, we have built a large water tank facility at our Biscayne Bay Campus that serves as the physical foundation for a virtual, video-game-like model.

The real tank is equipped with cameras that provide precise location data for the robot inside, helping us to create an accurate digital replica. We then used this virtual replica to run a wide variety of tests to help us research creative ways to improve how these robots operate. For example, our team used the digital twin to test different ways of controlling marine robots and develop virtual training environments for robotics education, including the integration of head-mounted displays, hand controllers and tracking gloves into a virtual interface. Overall, it’s a really useful facility that I think is just scratching the surface of its research potential.

How is AI improving how marine robots navigate?

AI allows robots to understand their environment as they operate. Instead of following pre-programmed paths, they use AI to predict where important data might be found, even with environmental noise or obstacles. They adjust their routes as they discover temperature hotspots, pollution sources or areas of concern.

Traditional systems relied on constant communication with a central computer to stay coordinated, but our AI-powered approach eliminates this dependency. Each robot operates independently, making its own decisions about where to go next based on what it's learning about water quality parameters like temperature, dissolved oxygen and turbidity. This autonomy is crucial in marine environments where communication can be spotty or lost entirely.

Under the guidance of my Ph.D. advisor, Leonardo Bobadilla, we developed methods in which robots communicate only with nearby teammates but still efficiently divide up large areas—like the lake behind the Green Library at FIU. They share their local findings using techniques to focus on the most useful patterns. This creates a "smart swarm" that can map how environmental parameters vary across space much more efficiently than traditional buoy methods.

How do you look back on the work you did at FIU?

I feel like FIU added significant value to my research career. Before coming here, I had my master's in electrical engineering from the University of Rio de Janeiro, did research at the University of Duisburg-Essen and the German Aerospace Center, and even spent time at Santa Clara University and Stanford.

Each of those experiences taught me something different about tackling complex problems in engineering and robotics—so when I arrived, I felt ready to make a genuine impact.

For example, I was proud that our work in environmental monitoring was supported in close partnership with the Florida Department of Environmental Protection. During my Ph.D. studies, we demonstrated ways to autonomously track hotspots of water quality parameters within a given region of interest. As a result, our research received three award nominations at IEEE ICRA 2025, the world’s premier conference on robotics and automation. I'm excited to see where this research goes from here.

It was also energizing to be part of a larger, multidisciplinary team. Our marine robotics research comprised multiple researchers at the College of Engineering and Computing, like Bobadilla and Gregory Reis, as well as our colleagues at the Institute of Environment.

What's next for you?

I am joining the Department of Mathematics and Computer Science at Providence College as an assistant professor this fall.

In my new role, I'll continue to advance robotics and machine learning research while also focusing on STEM education. I think it's a natural next step that builds on the work I did at FIU.