Engineers at Cornell University have developed a robotic system that behaves more like a flowing material than a traditional machine, allowing groups of small robots to move, reshape and adapt without centralized control.

The system, called the Cross-Link Collective, is made up of dozens of narrow robotic modules that can latch onto one another using weak Velcro connections. Individually, the robots move slowly and struggle in difficult terrain. But when linked together, they begin acting as a coordinated collective capable of navigating obstacles and inclines.

Researchers say the robots rely on “mechanical intelligence” instead of advanced computation or communication systems. Their physical interactions allow coordinated movement to emerge naturally as the modules continuously connect and disconnect while moving.

“Instead of relying on explicit computation and communication, the system shifts the intelligence into the shape of the robots and their physical interactions,” said corresponding author Kirstin Petersen, associate professor of electrical and computer engineering at Cornell.

Each robot measures roughly 200 millimeters long and 20 millimeters wide. A small internal motor drives the module to repeatedly switch between an “I” shape and a “U” shape, creating forces that push it forward across surfaces. Weak Velcro patches at both ends allow nearby modules to temporarily attach and detach as they move.

Robots move like fluids

When the robots form chains, their behavior changes dramatically. On sloped surfaces, the linked modules moved more reliably than single units, which frequently stalled depending on orientation.

In obstacle-filled environments, the chains behaved similarly to flowing materials. Connections formed to maintain group cohesion, but also broke apart when necessary to avoid getting stuck.

“It doesn’t matter if one module has a compromised battery or fails for other reasons,” said lead author Danna Ma. “The system stays functional because it can adapt. It is redundant and doesn’t depend on any single module.”

The researchers said the design allows the system to remain operational even when some robots stop working. Since no central controller directs movement, the collective can reorganize itself dynamically.

The team also demonstrated that adding a small amount of sensing improved coordination. When a robot becomes separated from the group, it emits an audible buzzing signal that prompts nearby modules to slow down, giving the isolated robot time to reconnect.

No central brain

“There is no centralized sensing or control,” Ma said. “Each module can infer when it has lost contact with the group by how much it’s being jostled and then use an audible buzz to slow down nearby modules while it catches up. It’s as simple as that.”

Researchers at Georgia Institute of Technology originally designed the robotic module. The Cornell team later refined the system through years of testing and statistical analysis to improve how effectively the robots connect and move together in large groups.

The Cross-Link Collective takes inspiration from active gels, materials whose molecular bonds constantly form and break while preserving overall structure. Researchers believe the work could help advance soft robotics and systems designed to operate in unpredictable real-world environments.

“It’s helpful for us to start thinking about what we can encode into the physics of a system itself, as robots are increasingly applied to real-world scenarios that are highly unreliable and dynamic,” Petersen said.

The study was published in the journal Science Robotics.