MTCOM connects USVs for coastal surveillance

MTCOM is a project coordinated by SMA-RTY, aimed at building an autonomous coastal surveillance system for the Port of La Spezia. At its core is a private 5G network that connects a fleet of USVs (Unmanned Surface Vehicles) to a centralized platform, enabling real-time data collection and situational awareness along the coast.

The project was developed together with a group of partners, each bringing a different piece of expertise to the table: Infocom, EdgeLab, Sitep, and the University of Genoa.

We already spoke with the University of Genoa about the physics of maritime radio signals and with Infocom about the orchestration layer powering the system, we now turn to EdgeLab.

We sat down, virtually, with Francesco Paolo Falotico and Affaf Momin, from EdgeLab, a La Spezia-based company specialising in autonomous marine vehicles, to talk about what they built, how it works, and why a surface drone that avoids boats is a bigger deal than it sounds.

Francesco: We are a company that builds autonomous marine systems, both underwater and on the surface. A human can control them remotely, or they can work fully autonomously depending on the mission. Most of what we do sits underwater, but for MTCOM we brought one of our surface vehicles into the picture.

Affaf: It is a catamaran, a small flat vessel with two hulls. It can be guided by an operator on the ground, or it can navigate on its own by following a set of waypoints that the user drops on a map. You basically draw the route and the vehicle follows it. It can even loop the route continuously, going back to point one after it reaches the last stop.

Affaf: Ha, actually yes, that is a pretty good way to put it. The user defines the path and the vehicle sticks to it. But just like a Roomba bumps into things and reroutes, our catamaran uses a stereo camera, a depth-sensing camera, to detect obstacles in real time.

If a boat, a buoy, or a pier suddenly shows up in the way, the vehicle avoids it automatically and then picks up its route again. That is especially important in coastal and harbour environments, which is exactly where MTCOM operates.

Affaf: Several things. There is a live video feed so the operator always knows what is in front of the vehicle. An echo sounder measures the depth of the water below. And we recently integrated a multiparametric probe that logs things like salinity, conductivity, and pH. All of that data is transmitted back to the ground station in real time via a long-range antenna on board.

Affaf: Exactly. The idea behind MTCOM is coastal water inspection: water quality, depth, navigation hazards. The catamaran gives you eyes and sensors in places that are hard to reach by boat, in a way that is repeatable and cost-effective.

Affaf: The catamaran platform itself existed. We built the first version back in 2023, originally to act as a surface companion for our underwater vehicles. But several key systems had to be developed from scratch for MTCOM.

The biggest one was the obstacle avoidance system. Processing images fast enough to dodge obstacles in real time, five to ten times per second, requires serious computational power that the vehicle did not have.

Affaf: We added a new computer on board with a GPU, a graphics processing unit, the same kind of chip used in video games, which is very efficient at running image-recognition algorithms quickly.

The clever part is that thanks to the Robot Operating System, which is the software framework we use, we did not have to tear apart the existing system. We connected the new computer to the old one via a single Ethernet cable inside the vehicle and the two worked together seamlessly.

Affaf: Sure. Tracking an underwater vehicle is tricky because GPS does not work underwater, so you have to rely on acoustic signals, basically location based on sound instead of light. And sound-based communication underwater is extremely slow: you are talking about one or two bytes per second.

So if you have a surface vehicle following your underwater drone, you can connect them with a fibre optic cable. The underwater vehicle sends its sonar images and data up to the surface vehicle, which then relays everything to the ground station over a normal antenna. The catamaran essentially acts as a bridge between what is happening underwater and the people watching on shore.

Affaf: Yes, Luca Maggiani from SMA-RTY sent us the dimensions and specifications of the antenna unit they wanted installed on the vehicle well ahead of the integration day. Based on that, we designed and built dedicated mounting fixtures and made sure there was sufficient space on the catamaran.

SMA-RTY also provided the electrical and network connections required, and we developed those on our side. So by the time we actually met in La Spezia, everything was already prepared. It was a matter of fitting the pieces together rather than figuring things out on the spot.

Francesco: Very smoothly. We had regular weekly calls throughout the project and my main point of contact was Luca. He has the mindset of a researcher: he asks the right questions and anticipates problems early.

On the day of integration we put everything together in a couple of hours and took the vehicle straight to the port for testing. No last-minute surprises, which in a project like this is something you really appreciate.

Francesco: It ticked two boxes at once.

First, we had this platform that we had not been fully exploiting, and it made sense to find a real application for it.

Second, MTCOM pushed us to develop expertise in machine learning and artificial intelligence for image recognition and obstacle avoidance. That is a domain we were already looking to move into, and the project gave us a concrete opportunity to do it.

Sometimes the right opportunity and the right technology just meet at the right moment, and for us that was MTCOM.

EdgeLab s.r.l. is based in La Spezia, Italy, and develops autonomous underwater and surface systems for maritime applications. MTCOM is led by SMA-RTY and focuses on enabling autonomous coastal monitoring using connected USVs.