FloatForm: MIT's tiny robot boats assemble floating structures

FloatForm, a swarm of small square robotic boats from MIT, assembles itself into larger, reconfigurable structures on the water and then breaks apart and reassembles on command. The open-access paper, published July 9, 2026 in Nature Communications, shows tabletop tests, failure rates, and simulations that scale the approach to 64 units.

How do FloatForm robots assemble and latch?

FloatForm robots are 21 centimeters square, each carrying its own thrusters, sensors, and a hidden magnetic latch, and they connect using an origami-inspired auxetic mechanism. A single servo motor drives a structure that pulls permanent magnets inward to release or pushes them outward to grab a neighbor across gaps of 10 to 15 centimeters; a 3D-printed gearbox locks the latch with the motor off so the mechanism consumes power only when changing states.

The boats move using four miniature thrusters arranged in an "X" for omnidirectional motion. Coordination combines distributed neighbor-to-neighbor position exchanges with a lightweight central planner that assigns final positions. The robots navigate toward those targets, avoid collisions, adapt to disturbances, and latch in parallel so the entire swarm can form a rigid lattice and then operate as a single vessel in collective transport.

How did FloatForm perform in tests and simulations?

In MIT tank experiments, a fleet of eight robots repeatedly gathered from random starting positions, latched into a rigid structure, broke apart, reassembled into a new configuration, and drove across the pool as one unit, with each run taking four to eight minutes. Across 10 trials, mission completion without human intervention occurred 90 percent of the time with four robots and 70 percent with eight.

The team reported that when failures occurred the architecture remained resilient: robots that lost their bearings could rejoin on their own, and units stuck in formation deadlocks learned to shake themselves free and retry. Simulations showed the coordination framework scaling smoothly to swarms of 64, because planning complexity depends on a robot's local neighbors rather than total swarm size.

Why it matters

FloatForm shifts the computational burden onto individual robots and away from a single central planner, avoiding single points of failure and reducing planning growth as the swarm expands. That enables parallel assembly and collective motion, which the team argues keeps overall assembly time from increasing significantly as the fleet grows. The system also gains physical stability by joining together, improving resilience to waves or currents compared with single small boats.

Those attributes point to practical uses beyond the lab: temporary bridges, floating markets, mobile stages, offshore inspection platforms, adaptive sensor networks, and reconfigurable docking or emergency-response platforms. The project builds on Roboat work in Amsterdam and the team explicitly frames FloatForm as a modular, city-facing way to use underutilized waterways.

What to watch

Watch for scaled prototypes that trade the lab's ultrasonic indoor positioning for GPS or vision-based sensing and for mechanical interlocking on larger boats, as the team notes those steps are needed to operate in disturbed real-world waters. The next milestone will be a field demonstration that reproduces the tank's collective transport and reconfiguration performance in an outdoor canal or harbor.

Daniela Rus, lead author Wei Wang, Alejandro Gonzalez-Garcia, Niklas Hagemann, and senior authors Carlo Ratti and Rus wrote the paper; the work was supported by the Amsterdam Institute for Advanced Metropolitan Solutions and other partners. Rus described FloatForm as opening "new possibilities for mobility, emergency response, public space, and infrastructure on water."