Many people think that the waterfront is at the edge of town. The MIT research team sees it as a powerful building block, similar to Lego.
Their new system, called “FloatForm,” is a collection of square robotic boats that assemble themselves into large structures on the water, separate, and reassemble into something new, all with minimal human guidance.
Each robot, roughly the size of a dinner plate at 21 centimeters square, is an independent vessel with thrusters, sensors, and magnetic latches. Together, they show a future where floating infrastructure may adapt: a temporary stadium after an emergency, a market on a canal, or a stage that emerges from a festival and dissolves when the crowd goes home.
“Our FloatForm projects envision a future where the ocean becomes a programmable extension of the city, where autonomous boats can organize themselves into bridges, platforms, and other useful structures when needed,” said Daniela Rus, Panasonic Professor of Electrical Engineering and Computer Science at MIT and director of MIT’s Computer Science and Artificial Intelligence Laboratory. “This type of distributed robotics opens up new opportunities for transportation, emergency response, public space and water infrastructure.”
“With FloatForm, we’re essentially turning static bodies of water into dynamic, programmable spaces,” said Wei Wang, lead author of the project’s new paper and a former MIT research scientist who now leads the Marine Robotics Lab at the University of Wisconsin in Madison. “Imagine an urban space where the public space is not fixed, but can automatically expand, contract, or reshape as needed.”
“We see it as building infrastructure on water, using a modular system to create one large system,” said Alejandro Gonzalez-Garcia, former MIT CSAIL and Senseable City Lab researcher. “If there is an emergency, you can build a new bridge to reduce traffic in the city. Or you can build floating markets and floating sections. If you want a livable city, you want to use water.”
The open access work, published today by Natural Communicationcomes from the labs of Rus and Carlo Ratti, professor of the practice of urban technology and planning at MIT and director of the Senseable City Lab, and grows from Roboat, their joint project with the Amsterdam Institute for Advanced Metropolitan Solutions that puts full-sized autonomous ships in Amsterdam’s canals. That canal once carried the goods of the city; today, they carry a lot of tourists.
“We explored whether canals could be used for waste collection, or transport, to relieve some pressures on waterways,” said Niklas Hagemann, an MIT graduate student in architecture, a CSAIL affiliate, and a former Senseable City Lab researcher who has worked on the project since its inception. “Urban areas are becoming more crowded, so can you expand public space into water that is currently underutilized?”
FloatForm shrinks that idea down to the scale of a tablet to answer a difficult question: How do you get more robots, and eventually thousands, to program themselves?
Ant raft studies
The team found their answer in biology. Fire ants survive the floods by connecting their bodies to life rafts, with no leader organizing the assembly. Each ant follows simple spatial rules, and a solid structure emerges.
“Each ant is an independent agent,” says Gonzalez-Garcia. “We wanted each robot to have its own power, the same way ants build a raft.”
Most existing self-assembling robots, in water and elsewhere, rely on a central computer that controls all movements. That approach is at risk of a single point of failure and it scales well: The scheduling equation balloons as more robots are added, and the swarm must meet in sequence, with many robots sitting idle while they wait their turn. FloatForm returns balance. A lightweight central planner moves slowly, giving each robot a place to complete the lattice, a level of geometric precision that distributed methods struggle to ensure. Everything else, including navigating to a target, avoiding collisions, and adapting to disturbances, is handled by the robots themselves, which communicate by exchanging positions with their nearest neighbors. All the waves move at once.
That similarity is what makes the work different. The programming complexity of the FloatForms method depends only on the neighborhood of the robot, not the absolute size of the float. “What we’re trying to do is have minimal intervention, and get them all moving at the same time,” Gonzalez-Garcia said.
In the experiments at MIT, an array of eight robots repeatedly assembled from random locations into a target position, attached to a rigid structure, split in half on command, reassembled in a new configuration, and then crossed the lake as one ship, each run lasting four to eight minutes. In that last method, called a transport cluster, the planner draws the route of the entire structure and each robot includes its contribution. “Every robot becomes an actuator,” explains Gonzalez-Garcia. The simulations showed the structure growing smoothly up to a mass of 64.
“The beauty of this method of decentralization is that the numbers don’t get fixed as the crowd grows,” said Wang. “Whether you’re working with eight boats or 80, all the ships coordinate and move at the same time. Because the overall coordination time does not increase significantly in principle, the system remains very dangerous.”
There is a physical benefit to sticking together, too. “Our boats become more stable by sticking together, like a raft of ants, when you have waves or currents,” Hagemann said.
Origami handshake
The robots connect via a connection mechanism that is completely hidden inside each skin. A single servo motor in the center drives the origami-inspired auxetic structure, a geometry that works equally in all directions at the same time, pulling the permanent magnets on all four sides inward to release them, or pushing them outward to hold the neighbor in spaces of 10 to 15 centimeters. The magnets are arranged in alternating polarities, so the boats click reliably on clean square laths.
The good part is what the machine doesn’t do: use (a lot of) energy. The 3D printed gearbox holds the latch in any situation when the engine is turned off. “It uses energy to connect and shut down, but between those states, it doesn’t use any energy,” Hagemann said. With infrastructure that can take hours to prepare, that’s important. “Because the robots are so small, you can have a very large battery,” added Gonzalez-Garcia. “If they use less energy to connect, they can use more to calculate, or to move.”
Getting there required some humble engineering. Four small thrusters arranged in an “X” give each robot omnidirectional movement, including turning in place, but they pack a lot of power relative to the robots’ small inertia, which made earlier prototypes wobble and prone to violent spins at low speeds. The team added stabilizing wings to increase hydrodynamic drag and tuned the controls to keep them stable for all robots, at this scale, which are never the same. The magnets presented their own problem: They held so well that turning off the wire sometimes required the robots to twist themselves free.
From the tank to the canal
In all 10 tests, the system completed its task without human intervention 90 percent of the time with four robots and 70 percent with eight. When things went wrong, the structure showed its resilience: A robot that lost its bearings for a moment was able to rejoin the structure itself, without stopping the whole mass, and robots stuck in the products of the structure learned to shake themselves and try again.
Going from a controlled indoor tank to a real canal or harbor will take more than confidence. Gonzalez-Garcia says: “There’s always a correlation between the size of the boat and the size of the disturbance it can handle. “These boats are very small, so in very disturbed waters, they can’t work.” Upscaling will mean tightening the latches, possibly by mechanically linking like the full-sized Roboat used, and trading the internal position of the lab with GPS or vision-based sensors. Helpfully, the linking algorithm was designed to be sensor-agnostic: exchange the senses, save the mind.
The team envisions applications beyond the city’s canals, from building temporary platforms for offshore testing and maintenance to flexible sensor networks for studying migratory species to adjustable docking stations for emergency response in hard-to-reach areas. There is also the potential for offshore and remote operations, from temporary construction bases to environmental monitoring and scientific expeditions.
And the geography is wide open. “Venice, the Netherlands, Belgium, the fjords and lakes of Norway, in fact any city with a river can use this opportunity,” said Gonzalez-Garcia. “This project uses areas where water is already important, but it also raises the question: Where else could water be used for something more?”
“This is an exciting step forward in realizing distributed behavior in water,” said University of Michigan Assistant Professor Steven Ceron, who was not involved in the research. “Assembly, reorganization, and coordinated movement are difficult enough on dry land, but implementing these behaviors in a highly distributed manner in water represents an additional big challenge, and this team has overcome it convincingly. By transferring the computing burden to the robots themselves, they have created a robust system that will in the near future be able to open space such as this robot to collect water for search, operation, environmental monitoring, and repairable marine infrastructure.”
Gonzalez-Garcia, Hagemann, and Wang wrote the paper with senior authors Ratti, who is also a professor at Politecnico di Milano, and Rus. Gonzalez-Garcia is affiliated with the MECO Research team at KU Leuven. The research was supported by a grant from the Amsterdam Institute for Advanced Metropolitan Solutions, with additional support from the University of Wisconsin at Madison. The team thanks MIT Sea Grant and Professor Michael Triantafyllou for providing the test tank.
