Octopus-inspired robot arm distributes intelligence to its suckers

A soft robot borrows the octopus' most useful trick

Underwater robotics has long been constrained by a familiar engineering model: rigid structures, centralized processors, and pre-programmed movements that work best when the environment is predictable. The ocean floor is not predictable. Currents change, visibility drops, and terrain shifts without warning. That is why a new soft robotic arm from the Italian Institute of Technology stands out. Instead of fighting ocean complexity with more top-down control, the system distributes perception and action through the arm itself.

The inspiration is the octopus, whose nervous system is notably decentralized. According to the researchers, roughly 60% of the animal's neurons are distributed across its eight arms, allowing local processing and reflexive action without waiting for instructions from a central brain. The IIT team translated that principle into a silicone-and-electronics architecture built for underwater exploration.

How the arm works

The robotic tentacle is 41 centimeters long and 4 centimeters in diameter at the base. It carries 10 artificial suckers that narrow toward the tip, echoing the layout of a real octopus arm. What makes the design distinctive is not just its softness, but its control philosophy. The arm relies on no cameras, no external computers, and no centralized command layer for basic contact response.

Each sucker contains three pairs of LEDs and phototransistors, optical components that measure reflected light. When an object touches a sucker, the silicone deforms and changes the reflection pattern. The system converts that change into three kinds of information: whether contact occurred, how much force was applied, and the angle from which the contact came.

The reported performance is precise. Sensitivity reaches about 400 millivolts per Newton, with a force margin of error of only 0.1 Newton. Directional accuracy is also tight, with a maximum error below 18 degrees and a mean around 8 degrees. Those numbers matter because they show the researchers are not merely imitating octopus anatomy for effect. They are building a sensing system that can support meaningful manipulation in uncertain environments.

Perception and action in the same place

Lead author Barbara Mazzolai described the design as one in which perception and action are integrated and distributed throughout the body. That sentence captures the larger significance of the project. In many robots, sensing happens in one place, computation in another, and movement in yet another. The octopus-inspired arm collapses those distinctions. A sucker does not simply report data upward. It interprets local contact and participates directly in the grip response.

That has practical advantages underwater. When communication delays or noisy conditions make centralized control cumbersome, local autonomy can improve responsiveness. A distributed system may also prove more resilient when unexpected contact occurs across irregular surfaces, delicate objects, or cluttered terrain.

Why this matters for ocean-floor exploration

The ocean is one of the clearest cases where biological inspiration can outperform more traditional robotics assumptions. A machine operating near the seabed may need to touch before it can see clearly, adapt without waiting for detailed instructions, and grip without damaging what it encounters. Soft bodies and local reflexes are well suited to those demands.

The IIT arm suggests a route toward that future. Rather than treating decentralized intelligence as a software feature layered on top of a rigid platform, the team embedded it into the mechanics of contact itself. The result is a robot that appears designed to react naturally when the environment refuses to cooperate.

The implications stretch beyond marine science. Any field that requires safe handling in unpredictable spaces could learn from this model. But underwater work is where the concept feels most immediately convincing, because that is exactly where centralized, vision-heavy robotics often struggles most.

The broader shift in robotics design

For years, high-performance robotics has often meant more sensors, more compute, and more explicit planning. This project points in a different direction. It argues that intelligence can be spread across the body, and that the shape and material of a robot can carry part of the computational burden. In other words, the control system is not only in code. It is in the structure.

That does not make centralized systems obsolete. It does suggest that the next generation of field robotics may be strongest when it blends central planning with local embodied intelligence. The octopus solved that problem long ago. Engineers are now catching up.

• The soft robotic arm is modeled on the octopus' decentralized nervous system.
• Each artificial sucker senses contact, force, and angle using LEDs and phototransistors.
• The system operates without cameras, external computers, or centralized control for local response.
• The design is aimed at better performance in unpredictable underwater environments.

This article is based on reporting by New Atlas. Read the original article.