Soft Robotics

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI explores the boundary between human and machine competence. Soft robotics is where that boundary gets physically intimate: machines that can be worn, that can work in direct contact with the body, that are inherently safe around people because their material properties absorb collision rather than transmitting it. The soft robotic exoskeleton that helps a laborer lift, the wearable system that monitors a patient’s condition, the surgical assistant that steadies a hand: these are the literal superpowers Rus imagines—the machine not as rival but as second skin.

Soft robotics also provides the most concrete demonstration of the morphological-computation principle that runs through physical intelligence: intelligence distributed through the body rather than concentrated in the processor. This is a direct challenge to the screen-centric conception of AI as something that happens in data centers. In soft robotics, the material is the computation, and the body is not a passive vessel for intelligence but a participant in it.

Origin

The field emerged as a distinct discipline in the 2000s and 2010s, drawing on silicone molding, pneumatic actuation, and 3D printing to create robots without the rigid links and precise joints of conventional systems. Rus and colleagues at MIT were central to its development, including the SoFi robotic fish—a soft, silicone-tailed robot that swam with real fish along a Pacific coral reef in 2018—and research into soft grippers, wearable robotic systems, and 3D-printed robots that integrate mechanical and electronic components in a single fabrication step.

The intellectual roots are in biology and in the phenomenological recognition that living bodies do computation through their shape and material properties, not only through their nervous systems. An octopus’s arm is semi-autonomous; it can grasp and manipulate with relatively little direction from the central brain because the local neural tissue and the mechanical properties of the tentacle itself handle much of the control. Rus’s argument is that artificial machines can and should be designed the same way.

Key Ideas

Morphological Computation. A soft body that passively conforms to its environment performs computation that a rigid body would require an explicit controller to achieve. The intelligence is distributed into the material. This principle allows soft systems to be simpler, more efficient, and more robust than rigid systems handling the same tasks in unstructured environments.

Safety Through Compliance. A rigid machine with the strength to be useful in physical tasks is a machine that can injure a person in its vicinity. A soft machine, because its structure absorbs and distributes force rather than concentrating it, is intrinsically safer to work alongside. This compliance—the property that makes the machine weaker in some respects—is also the property that opens the full range of human environments to it.

Daniela Rus

The Body as Second Skin. Rus envisions soft robotic clothing: lightweight wearable systems that could add strength to a laborer, monitor a patient’s vital signs, provide independence to an aging person, or give expert-level feedback to a surgeon or athlete. This is embodied amplification of human capability rather than its replacement—the machine woven so closely into human life that the boundary between person and tool begins to dissolve.

Biological Inspiration. Soft robotics consistently returns to biological models not as metaphors but as engineering blueprints: the silicone fish tail that mimics the undulation of a real fish to move silently through water, the gripper modeled on an octopus’s arm, the wearable system that takes its cues from the mechanics of muscle. Nature has solved the problem of acting effectively in the world using soft, compliant structures, and the field’s working assumption is that these solutions contain architectural lessons applicable to machines.