If I Only Had a Brain

We are asking robots to do more complex things all the time, like drive our cars for us, assist surgeons with delicate procedures in the operating room, and clean our homes. To make these things possible, robots rely on central processing systems — analogous to brains — to sense the world around them and respond appropriately to carry out their tasks. As you might expect, the more complex a task is, the more powerful the computing hardware must be.

That is no problem for some applications, like a self-driving vehicle. But when the job demands that a robot be small or soft, traditional computing hardware just can’t cut it. With their relatively large sizes and rigid components, these systems are impractical for use in applications where tiny sizes or flexibility are required.

An interesting workaround may be on the horizon, however. A group led by researchers at the University of Oxford in the UK has reimagined what a robot control system might look like. Their solution needs neither compute power nor electronics for operation. Instead, they built sensing and decision-making capabilities directly into the physical structure of their soft robots. When air pressure is applied to these structures, they can act as actuators, sensors, and even logic gates.

The team calls their creations "fluidic robots." Unlike traditional robots that depend on electronics, motors, or computer chips, these soft robots achieve complex, coordinated motion purely through mechanical interactions between air pressure, flexible materials, and their surrounding environment.

Each robot is built from identical modular fluidic units, each just a few centimeters across. These units are the building blocks of the system, comparable to LEGO pieces that can be connected in different ways to create robots capable of unique behaviors. Depending on how a unit is configured, it can perform as a muscle-like actuator, a pressure or contact sensor, or a valve that controls airflow — or even all three at once.

When air pressure is continuously applied, the units deform and recover rhythmically, creating self-sustained oscillations. By linking several units together, the motion of one influences the others through the shared robot body and the ground beneath it. This coupling leads to spontaneous synchronization, where multiple limbs begin moving in unison — a phenomenon the researchers explain using the Kuramoto model, a mathematical framework for understanding how oscillators synchronize in nature.

The result is emergent behavior. The robots coordinate their movements automatically without any traditional programming or electronic control. For example, the team demonstrated a crawling robot that could sense the edge of a table and stop before falling, and a shaker robot that sorted beads by tilting a platform back and forth. Each function arose purely from the configuration of air channels and flexible components, not from digital computation.

While the current prototypes are tabletop-sized, the design principles are scale-independent. In the future, these air-powered robots could be adapted for use in environments where electronics fail or power is scarce, such as in deep-sea exploration, disaster response, or even space missions.