Laurent Giraid has spent his career at the intersection of artificial intelligence and physical systems, focusing on how distributed intelligence can solve complex logistical problems. Recently, he has been exploring the transformative potential of “FloatForm,” a revolutionary modular robotics system developed at MIT that redefines urban architecture. This project envisions water surfaces not just as edges of a city, but as dynamic, programmable spaces where autonomous, dinner-plate-sized boats can self-organize into bridges, markets, or emergency platforms. By moving away from centralized control and embracing the biological logic of social insects, this technology suggests a future where our infrastructure is as fluid and adaptive as the water it floats upon. In this conversation, we explore how these tiny vessels navigate the challenges of hydrodynamics, energy efficiency, and the complex math of swarm coordination to turn underutilized canals into the construction sites of tomorrow.
How does mimicking the local interaction rules of fire ants improve the scalability and resilience of these robotic swarms compared to traditional centralized systems?
When you look at a fire ant colony during a flood, you see a living raft where no single leader is giving orders, yet the structure remains perfectly intact. We took that exact biological inspiration for the FloatForm system to overcome the “planning math” bottleneck that usually kills large-scale robot projects. In a traditional centralized setup, a single computer has to calculate every single move for every single boat, which means as you go from eight robots to 80, the system eventually chokes on its own complexity. By using local rules, each of our 21-centimeter robots only needs to worry about its immediate neighbors, which allows the entire swarm to move in parallel rather than waiting their turn. This parallelism is the secret sauce; in our simulations, we saw the system scale smoothly to swarms of 64 robots without the computation getting bogged down. If one robot loses its way or experiences a mechanical hiccup, the rest of the group doesn’t just stop and wait; they continue their mission while the “lost” unit uses its own sensors to rejoin the lattice once it regains its bearings.
The physical connection between these boats seems vital to their stability. Could you explain the engineering behind the origami-inspired latching mechanism and why energy efficiency was such a priority?
The latching mechanism is really a feat of elegant, minimalist engineering that lives entirely inside each 21-centimeter hull. We used a single servo motor to drive an origami-inspired auxetic structure—a specific geometry that can contract or expand uniformly in all directions at once. When the motor engages, it pulls permanent magnets inward to release a neighbor, or pushes them outward to grab onto another boat across gaps of 10 to 15 centimeters. The brilliance of this design is that we used a 3D-printed gearbox that holds the latch in place even when the motor is switched off. This means the robots consume zero energy to stay connected, which is crucial because when you are working with a vessel the size of a dinner plate, you can only fit a very small battery inside. By saving energy on the physical bond, we can redirect that precious power toward the computation and the four thrusters needed to keep the structure stable against the rhythmic pulse of moving water.
What were the specific hydrodynamic challenges your team faced when trying to prevent these small, lightweight robots from spinning out of control?
Early on, the prototypes were incredibly “twitchy” because the four miniature thrusters, arranged in an “X” configuration, packed a huge amount of force relative to the robots’ tiny inertia. At low speeds, even a small burst of power could send a boat into an aggressive, frustrating spin that made precise docking nearly impossible. To solve this, we had to add stabilizing fins to the bottom of the hulls to increase hydrodynamic drag, which essentially acts like a set of training wheels in the water. We also had to spend a lot of time tuning the controllers to account for the fact that, at this miniature scale, no two robots are ever truly identical in how they move. Once they latch together, however, the physics change for the better; the individual “twitchiness” disappears and the collective structure becomes much more stable against waves and currents, much like how the individual fire ants become a resilient, buoyant mass.
The FloatForm system is described as having a “lightweight” central planner. How do you balance that central oversight with the autonomous decision-making of the individual boats?
We use the central planner very sparingly, almost like a choreographer who only steps in to give the final positions of a dance. Its primary job is to assign each robot a specific coordinate in the final lattice to ensure we get that geometric precision that purely distributed systems often struggle to achieve. Once those “target seats” are assigned, the central computer steps back and the robots take over, navigating toward their targets while autonomously avoiding collisions with their peers. This hybrid approach allows us to have the best of both worlds: the high-level organization of a blueprint and the real-time agility of an independent agent. During our tests, we saw this in action as a fleet of eight robots would gather from random spots, latch into a rigid structure, and then transition into “collective transport” mode where every robot becomes an active actuator moving the whole unit as one vessel.
Could you walk us through the results of your pool experiments and what they revealed about the system’s current success rates?
In our indoor tank at MIT, we ran dozens of trials where we asked the robots to assemble, break apart, and reconfigure into different shapes, usually taking between four to eight minutes per run. For a small fleet of four robots, the system was incredibly reliable, completing its mission without any human intervention 90 percent of the time. As we doubled the swarm size to eight robots, the success rate dipped to 70 percent, largely because the physical interactions and potential for “deadlocks” become more frequent as the crowd gets denser. We actually found it fascinating to watch how the robots handled these failures; when they got stuck in a formation deadlock, they would actually “shake” themselves free and retry the maneuver until the magnets clicked into place. These experiments proved that the coordination logic works, but they also showed us that moving to the “wild” waters of a canal will require even more robust sensors to handle the unpredictable environmental noise.
Beyond simple bridges, what is the broader vision for how this technology could transform the “underutilized” water surfaces in cities like Amsterdam or Venice?
We see the waterfront as a dynamic, programmable extension of the city rather than a fixed boundary. Imagine a canal in Amsterdam where, on a Friday night, a series of these robots autonomously gather to form a floating stage for a concert, and then by Saturday morning, they reconfigure into a temporary bridge to help move foot traffic during a busy festival. In more urgent scenarios, these swarms could be deployed as adaptive sensor networks to study migratory species or as rapidly deployable platforms for emergency response in areas that are hard to reach by land. By shifting the heavy lifting of infrastructure from static concrete to modular, autonomous robots, we are essentially turning the water into a flexible public square. This approach allows a city to “expand” or “contract” its public space on demand, making urban environments far more livable and responsive to the needs of the people who inhabit them.
What is your forecast for the future of reconfigurable marine infrastructure over the next decade?
In the next ten years, I expect we will move away from the idea of “docks” and “bridges” as permanent, heavy structures and toward a model of “infrastructure as a service” that can be summoned with a line of code. We will likely see these small-scale modular systems scale up significantly, trading our current indoor positioning for GPS and vision-based sensing that can withstand the glare and chop of the open ocean. I anticipate that major port cities will begin integrating these autonomous swarms into their daily waste collection and transit loops, offloading the immense stress currently placed on our road networks back onto the water. Eventually, this will culminate in fully autonomous, self-healing platforms that can maintain themselves during offshore inspections or scientific expeditions in the fjords of Norway or the lakes of the Midwest, effectively treating the water surface as a new, high-tech layer of the global urban fabric.
