Skip to main content

Science and Society Hard problems for soft robots

Machines have changed how we think about data. Will they make us think differently about matter, ask ROX MIDDLETON, LIAM SHAW and JOEL HELLEWELL

ROBOTS are here. They run our lives, they do our jobs, they find our soulmates, they persuade us what to listen to, how to spend our time and money, they teach us, they connect us to other people.

Your phone is a robot, your checkout is a robot, your smart speaker is a robot, your computer is a robot, your electronics are robots, using the internet means using robots to use other robots.

Perhaps calling all of these technologies “robots” is unnecessarily simplifying. It’s not how we imagined the future of machinery when the first automatons of the industrial revolution tumbled clanking and steaming into factories, or how robots were imagined in most visions of the future in sci-fi comics and films. And yet this is what we have made.

The “smart speaker” is perhaps the latest iteration of the most clearly recognisable robot. It has been designed for friendly interaction on a daily basis and to respond via speech to a wide variety of requests. The incredible success of robot intelligence has been radically re-imagined by access to the vast data storage, technologies and surveillance of the internet age.

The previously unimaginable reservoirs of data and adaptable algorithms, technologies of long-distance immediate communication and rewritable data storage have made robot intelligence currently nothing like we thought it would be. For better and worse, society has been revolutionised by the internet.

The striking thing about all the complex robots we now regularly live alongside, is that unlike imagined robots of the past, all the most complex robots are reduced to physically inert boxes. Our robots mostly don’t have bodies, we accept them as artificial brains in boxes.

That is to say, unlike the information revolution, which changed the way we understand robot intelligence, there has been no corresponding radical shift in the way that we can imagine robot embodiment. Where there have been efforts to replace workers with machines, the work-robots look like the manually operated vehicles, production lines, tools and tills they replace.

Engineering bodies is really hard. One of the hardest problems is applying exactly the right amount of pressure to grip and hold any object. Although we barely have to think about it ourselves, it’s really hard to program a hand that can deal with lightbulbs, strawberries, spanners and babies.

Handling any one of these presents a delicate problem of dexterity: imagine trying to design a robotic hand to manage them all. The way that robotics engineers tackle this is by using feedback circuits of sensors and actuators (the moving parts), constantly checking that the pressure for the target object is appropriate and regulating the movement.

This seems like a good analogy to our own manipulation, as we use nerves to sense and regulate the use of our muscles. (Correspondingly, nerve damage can cause debilitating difficulties in dexterity for people.)

Engineering improvement therefore involves speeding up sensing and feedback loops to make sure grippers react to what they’re holding. However, there is another approach, which its proponents hope will entirely change the way we think about robots. This is the realm of soft robotics.

Soft robotics covers an enormous range of different ideas, all to do with replacing cogs and levers with soft materials that move themselves. Starting to think about soft robotics starts off with considering rubber. Rubber has extraordinary electrical properties, and can be induced to expand vastly when attached to a voltage (and vice versa in order to generate a voltage from stretching).

Rubber properties such as stiffness and strength can be changed over wide ranges. It’s a prime candidate for soft robotics, because layering rubbers with different expansion properties is a way directly to devise all sorts of moving parts with complex shapes, like speakers and grippers. These movements are quite simply induced by voltage, making rubber easy to put into existing electric circuits.

Applied research has for example been done on how to build supportive exoskeletons for people with difficulty standing unaided by adding electrified rubber into clothing. Unfortunately these technologies are still far from widespread use. One real challenge is that the voltages needed to induce miraculous changes in rubber are enormous.

Voltage itself doesn’t cause damage if it can be entirely insulated, but with any failure in insulation the current that flows because of the voltage would easily kill you. Research is ongoing into materials that might be able to convert the extraordinary properties of rubber-like stuff into use at lower voltages.

Another possible feature for soft robots is in the form of gels and sponges that contain ionic liquids. Ionic liquids have charged atoms in them that move through the liquid when attached to a voltage, like electrons move through a wire.

By immersing a porous spongy solid in an ionic liquid, charge can pass through it, and therefore deliver electric charge through the material. This current also interacts directly with the shapes and pressures the material experiences. Although counterintuitive in circuits, the use of charged liquids to carry signals is what powers electrical signalling between biological cells.

By changing the way that charge is carried, artificial electric circuits can be re-imagined. Sponges and gels which incorporate liquids could be invented that reconfigure themselves under electrochemical stimuli, with potential to make a truly shapeshifting flubber-like goo.

In the most extreme cases soft-robotic movement can happen without the need for electrical circuits at all. Here, soft robotics seeks to find materials that respond to their environments directly, eliminating the need for a sensor and feedback loop.

Scientists are working on inventing materials that react to light, heat, chemicals or touch by moving, expanding or stiffening. Some of these are shape-memory materials, which are pre-programmed with a specific shape, which can be triggered again after deformation to other stable shapes.

Others are made of molecules designed from the bottom up to respond to energetic signals. What about a solar panel that tilted with the sun, without a motor or moving parts? Still others rely on micro — or macro — structure and use complex origami geometries to convert passive material changes into other movements. Think of a glove that gripped you back when you held it. By understanding materials as potentially sensory and self-moulding, we open a door to a new paradigm of robotics.

These materials are still in labs and there is still a long way to go before they become useful machines, but in doing so they will change how we think about robots and even matter itself.


We're a reader-owned co-operative, which means you can become part of the paper too by buying shares in the People’s Press Printing Society.

Become a supporter

Fighting fund

You've Raised:£ 15,468
We need:£ 2,532
2 Days remaining
Donate today