The robot finger bends inward with force. Although ‘robot finger’ is a big word for this curly plastic strip in the soft robot laboratory at the University of Twente. It’s more like a scouring pad in a vacuum-packed sandwich bag. A hand is not on it, let alone a robot. And yet, it is easy to imagine that one day a whole arm of this plastic will have to be made, which can perform tasks that have to be done with a soft hand, such as picking fruit.
When you think of robots, you might think of Disney’s Wall-E, cool Transformers or C3PO and R2D2 from Star wars† In short, humans made of steel. In fact, most robots are hardworking workers who perform their work in sheltered environments. At the car factory or laboratory e.g. Traditional robots are great for work that requires a lot of power, precision or speed. The mail sorting machine is a robot, just like the coffee machine is.
But the desire to involve robots in the real world is growing. To process food, to research in hard-to-reach places and to provide help in, around and next to the human body. It requires skills that traditional robots are not good at. The real world is more complex than the secure cage. It requires adaptability and a gentle nature.
A new class of robots is therefore on the way: soft robots. Made of soft, moving materials, often plastic, rubber or textiles, and without the hinges and motors that are so characteristic of traditional robots. The soft robots are often inspired by nature, where adaptability and softness are often combined with strength and precision. They move on air, in a vacuum, with current, thanks to electrostatic attraction or shape memory.
It is also best to program a traditional robot hand to gently close around a fruit. But it involves many sensors and small hinge movements. You breathe or vacuum a soft robot hand, and that’s it. The movement is subtle in itself, the material does the work.
At least that’s the idea. Soft robots with full ‘bodies’ and full functionality can so far be counted on two hands. Researchers are still working on basic questions: how do we make something move, how do we make it feel? And even: what are we going to make of what materials do what I want?
Welcoming and user-friendly
These issues have been worked on for about ten years now. It started slowly. At the beginning of the decade, magazines published several dozen surveys a year. It has grown to more than two thousand published studies by 2021, shows a brief search in the scientific search engine Scopus.
“Strictly speaking, the field is already older,” says Ali Sadeghi, head of the soft robot laboratory at the University of Twente. “In Japan, researchers in the 1980s wrote about a new class of robots that would be more compatible and user-friendly than traditional robots. They also designed some examples, but it did not work.” Important reason: the shapes they had in mind could not yet be made properly from soft materials.
It was only about ten years ago that more became possible. “A lot of scientists back then were working on electroactive polymers, plastics that deform when they get power,” Sadeghi says. “The patent for 3D printers also expired in 2009. Until then, such a device cost ten to twenty thousand euros, and then it was suddenly available for about two hundred euros. Design with new materials and shapes became much more accessible as a result.” At the same time, interest in bio-inspired designs grew.
In the Netherlands, TU Delft, Eindhoven University of Technology, Wageningen University and the University of Twente established the Dutch Soft Robotics consortium in 2019 to bring together and strengthen the research in soft robotics that took place at various Dutch universities. Sadeghi’s laboratory in Twente, where the spongy plastic robot finger is demonstrated, has been around for over a year now. The laboratory has the coziness of a cluttered kitchen filled with equipment. The research here is based on 3D printers, five of them, placed on a long table against the back wall.
The 3D printer is so useful because any shape imaginable can come out of it. A mold is built up in thin layers. This is usually done with plastic wire that comes off a roll and hardens when it comes out of the nozzle. In the soft robot laboratory, they will work with softer materials. It is possible, but not just like that. Anyone who comes closer will see that almost all printers have been tampered with to make this happen. An extra reservoir is glued to one with tape, to another another nozzle is added. The way the plastic is brought to the nozzle has also been adjusted in a number of printers.
“We make that spongy thing that I just bent with thermoplastic rubber balls,” says Nick Willemstein. He is a PhD student in biomechanical engineering at the University of Twente and is researching new possibilities for printers and materials. He shakes a few bullets from a glass jar in his hand. They are reminiscent of the balls you find in a shoebox to absorb moisture, but softer. “Because they are soft, it is difficult to get them through the nozzle. You can not push them through as you can with a hard plastic wire. Therefore, we put a kind of screw over the nozzle. The grooves contain the balls, which are thus forced against the nozzle. ”
Willemstein is the first to print a soft robot finger in this sponge-like structure, which consists of small plastic circles that lie above and next to each other. He originally printed simple squares to test stiffness at different densities. But there are a number of examples on the table, all of which have different shapes, each with a rectangular base with thickenings on top. In one, the thickenings are just across the width, with the other in diagonals. They all make a different move as Willemstein puts a vacuum on them using a plastic bag. Straight up like a finger, diagonal gives a twist like a wobble.
In soft robot technology, the following applies: the shape is the movement. Unlike traditional robot technology, where hinges are often a part of the body where you can set rotation and program movements, the movement with soft materials is embedded in the material and designed. Willemstein’s sponge can either make a twist or roll it up.
The fusion of form and function also means that knowledge from many fields is gathered in a soft robot. Traditional robot technology is based on engineering and computer science, while soft robotics also requires knowledge from mechanics, materials science and even biology.
The visit to the laboratory would actually have taken place months earlier. Sadeghi and Willemstein were hoping to show the art of a new printer designed by them, but they are still waiting for parts. They want to print two types of soft plastic at the same time so that the properties of the materials can be easily combined. One plastic is, for example, a soft base material, the other is soft and conductive. Willemstein must now always clean the nozzle in between. A time consuming job and yet the result will not be as nice as desired.
“We use conductive material to make the robot feel,” says Willemstein. The advantage of soft parts is that they bend nicely, but if you do not quite know where the bend is and how strong the bend is, you can not do much with the robot. “If he can feel, he can only interact with the things he picks up.”
Willemstein places a version of his sponge between two metal plates that also contain the conductive plastic and allows a current to flow. The harder he presses the handle upstairs, the less resistance the current encounters on its way from one metal plate to another. On the screen of his laptop, we see a line move as he presses and releases. “This way you can measure how strongly the material is deformed,” says Willemstein. It is not yet possible to say exactly where the distortion is. They are still working on it.
Colors on the inside
There are several techniques for measuring distortion. With light in combination with color, it is also possible to find out where the distortion is. At TU Delft, for example, a hollow robot hand with colors on the inside was developed last year. Put a candle on it, measure the color of the reflective light and you know where it bends. Other researchers did pretty much the same thing, but with stained glass fibers over the back of the hand. And then there are air pressures, magnetic fields and all kinds of sensitive foils and coatings that can be used for measurements.
The laboratory in Twente wants to expand the possibilities for the 3D printer further in the coming years. Take silicone rubber, a popular material for making soft robots. This is usually poured into a mold using molds. The rubber can dry gently in a mold. But it is a tricky technique where unwanted bubbles, for example, can destroy the robot. Printing prevents this, and it opens up other possibilities in terms of form.
But anyone who has ever pressed syrup to a fine shape on a pancake knows: liquid material runs out. Sadeghi and his students therefore experiment with the right mix of soft rubbers and look for the perfect distance between the reservoir and the nozzle so that the rubber has already solidified somewhat and no longer sinks enormously once it has been printed.
Although the first soft robots that look like an ‘end product’ are already there, most scientists are still in this creative process so far. In addition to the scam, simplicity entices. No more hinges that motor and jerkily perform their work. Inflate, vacuum or turn it on and you are done. Subtlety guaranteed.
“I’m hoping that in a few years, 3D printers will be successful enough to print soft, movable robot shutters in any shape you want,” Sadeghi said. “The dream is that in no time we can print something that looks like an animal. I also see all kinds of applications such as portable, for people with physical disabilities. And at home: A sofa that adapts perfectly to the body of the person sitting on it is really no fiction. But first we need to get a handle on the movement. No soft robot without controlled movement. ”