Tag Archives: soft robots

Soft Robots: Has Their Time Come?


image: University of Toronto

When I first saw a room full of soft robots I was a little freaked out. A bunch of guys standing over what looked like colostomy bags or oversized centipedes as they navigated an obstacle course. It was hard to see the potential in it, but when I took a closer look I realised soft robots were the missing link in the computational evolutionary chain. The biggest problem we have with robots — whether they’re on industrial production lines, in homes, offices, airports — they’re too rigid. They’re parodies, robots that look like robots, instead of assuming the contours, movements, materials of nature, which is where they should be heading. Think adaptable, flexible, dexterous, able to go places rigid robots can’t — and safer. Hence soft robots. (Here’s the piece I wrote more than two years ago for Reuters) 

There’s been huge progress in the limbs, skin, movement and energy sources of soft robots. But there’s still a long way to go. For example, soft robots by definition have soft outer layers, and those layers need to stretch and adapt like the outer layers of the animals they loosely mimic. They also need to be energy-sensitive, since soft robots are (usually) self-contained, untethered things that rely on portable electronics for power — if that. 

Researchers at Carnegie Mellon University have developed such a material that can adapt its shape in response to its environment — shape-morphing, in the words of the researchers. Carmel Majidi, an associate professor of mechanical engineering who directs the Soft Machines Lab at Carnegie Mellon is quoted as saying:  “Just like a human recoils when touching something hot or sharp, the material senses, processes, and responds to its environment without any external hardware. Because it has neural-like electrical pathways, it is one step closer to artificial nervous tissue.” (You can see some videos of the material in action here.) 

The composite, which is made up of liquid crystal elastomers (LCEs, a type of the LCD you see in flat panel displays but linked together like rubber) and liquid metal gallium indium, is also resilient and, to an extent, self healing:

“We observed both electrical self-healing and damage detection capabilities for this composite, but the damage detection went one step further than previous liquid metal composites,” explained Michael Ford, a postdoctoral research associate in the Soft Machines Lab and the lead author of the study. “Since the damage creates new conductive traces that can activate shape-morphing, the composite uniquely responds to damage.”

The researchers believe the material could be used in healthcare, clothing, wearable computing, assistance devices and robots, and space travel.

Another problem with soft robotics relying on soft actuators is that they tend to be bulky. Soft robots need to move and do stuff, and often this is done by pumping air or fluids through chambers. One — the colostomy bag lookalike — moved around in this way, and it was effective but took up a lot of space — a pump or something like it is usually required, which keeps them tethered and unwieldy. Researchers at UC San Diego reckon they have a solution in creating soft actuators that are controlled not by air or fluid but by electricity. (h/t Nanowerk)

If that sounds like a step backward, the point here is less that they’re using electricity, but are using material that is used for artificial muscles in robots, the LCEs I mentioned above. As with the CMU researchers, the UC San Diego team focused on how LCEs change shape, move and contract in response to stimuli such as heat or electricity. They sandwiched three heating wires between two thin films of LCE. The material was then rolled into a tube, pre-stretched and exposed to UV light. Each heating wire can be controlled independently to make the tube bend in six different directions, as well as contracting. 

The researchers built an untethered, walking robot using four actuators as legs. This robot is powered by a small lithium/polymer battery on board. They also built a soft gripper using three actuators as fingers. The thing was slow — each leg takes about 30 seconds to bend and contract, but they’re working on ways of speeding it up. 

Movement of soft robotics is a challenge. There’s lots of biomimicry involved, where researchers seek inspiration from land and sea creatures. Researchers at the University of Toronto have created a miniature robot that can crawl like an inchworm. This uses electrothermal actuators (ETAs), devices made of specialized polymers that can be programmed to physically respond to electrical or thermal change. A robotic inchworm in itself isn’t that novel— I saw one up the road at NUS here in Singapore a couple of years ago  — but the Toronto folk say theirs is different largely because it’s more efficient. And, I’d have to say, more like a real inchworm. They say their approach can be applied to other movements, including the wings of a butterfly. 

Their goal: to see it in clothing. “We’re working to apply this material to garments. These garments would compress or release based on body temperature, which could be therapeutic to athletes,” says Hani Naguib director of the Toronto Institute of Advanced Manufacturing, and the manufacturing robotics lead of U of T’s Robotics Institute. The team is also studying whether smart garments could be beneficial for spinal cord injuries.

There are other announcements — all of them in the past few weeks — that suggest major progress in this field: 

  • A Florida State University research team has developed methods to manipulate polymers in a way that changes their fundamental structure (think caterpillar turning into butterfly);
  • Soft robots could get smarter at solving everyday tasks after a team from MIT and Tsinghua University have developed a “soft finger” with embedded cameras and deep learning methods to allow the robot to better understand and manipulate their position, environment and items in it; 
  • Researchers from Linköping University in Sweden have come up with a way to fabricate soft microrobots from a single design process, hopefully making it easier to use soft robots in minimally invasive surgery or drug delivery. The researchers, intriguingly, say they “are now working on soft robots that function in air.” 

Look, I don’t think these things are going to find their way out of the lab any time soon. But clearly serious headway is being made. And in the end if it’s seen as commercially viable we’ll see big players get involved. So far there are a few players: Breeze Automation of San Francisco (a piece on them here from TechCrunch), and Fusion Fund, which says it’s interested in funding entrepreneurs using soft robots for “task automation beyond the capacities of current robotics technology.” It see soft robotics beyond industrial manufacturing — and I think they’re probably right. Soft robots will thrive in places either humans (and other devices) can’t get to — think search and rescue, like a Thai cave to reach stranded boys, or in interacting with humans safely and in an engaging way (a robot that can hug or catch a falling person, anyone?) and in miniature — hard to reach places inside an engine, inside a blood vessel, or in water. 

Going Soft On Robots

Snake on a plane! Don’t panic, it’s probably just a (soft) robot

My piece on soft material robots for Reuters. Original story here: I’ve added links in this piece.

SINGAPORE (Reuters) – Robots are getting softer.

Borrowing from nature, some machines now have arms that curl and grip like an octopus, others wriggle their way inside an airplane engine or forage underwater to create their own energy.

This is technology that challenges how we think of, and interact with, the robots of the not-too-distant future.

Robots are big business: by 2020, the industry will have more than doubled to $188 billion, predicts IDC, a consultancy. But there’s still a lot that today’s models can’t do, partly because they are mostly made of rigid metal or plastic.

Softer, lighter and less reliant on external power, future robots could interact more safely and predictably with humans, go where humans can’t, and do some of the robotic jobs that other robots still can’t manage.

A recent academic conference in Singapore showcased the latest advances in soft robotics, highlighting how far they are moving away from what we see as traditional robots.

“The theme here,” says Nikolaus Correll of Colorado University, “is a departure from gears, joints and links.”

One robot on display was made of origami paper; another resembled a rolling colostomy bag. They are more likely to move via muscles that expand and contract through heat or hydraulics than by electricity. Some combine sensing and movement into the same component – just as our fingertips react to touch without needing our brain to make a decision.

These ideas are already escaping from the lab.


Rolls-Royce, for example, is testing a snake-like robot that can worm its way inside an aircraft engine mounted on the wing, saving the days it can take to remove the engine, inspect it and put it back.

Of all the technologies Rolls-Royce is exploring to solve this bottleneck, “this is the killer one,” says Oliver Walker-Jones, head of communications.

The snake, says its creator, Arnau Garriga Casanovas, is made largely of pressurized silicone chambers, allowing the controller to propel and bend it through the engine with bursts of air. Using soft materials, he says, means it can be small and agile.

For now, much of the commercial action for softer robots is in logistics, replacing production-line jobs that can’t yet be handled by hard robots.

Food preparation companies and growers like Blue Apron, Plated and HelloFresh already use soft robotics for handling produce, says Mike Rocky, of recruiter PrincetonOne.

“This is an area robots traditionally can’t do, but where (soft robots) are on the cusp of being able to,” said Nathan Wrench of Cambridge Consultants.


Investors are excited, says Leif Jentoft, co-founder of RightHand Robotics, because it addresses a major pain point in the logistics industry. “Ecommerce is growing rapidly and warehouses are struggling to find enough labor, especially in remote areas where warehouses tend to be located.”

Some hope to ditch the idea that robots need hands. German automation company Festo and China’s Beihang University have built a prototype OctopusGripper, which has a pneumatic tentacle made of silicone that gently wraps itself around an object, while air is pumped in or out of suction cups to grasp it.

OctopusGripper (photo: festo.com)

A soft robot fish from China’s Zhejiang University swims by ditching the usual rigid motors and propellers for an artificial muscle which flexes. It’s lifelike enough, says creator Tiefeng Li, to fool other fish into embracing it as one of their own, and is being tested to explore or monitor water salinity.

And Bristol University in the UK is working on underwater robots that generate electrical energy by foraging for biomatter to feed a chain of microbial fuel-cell stomachs. Hemma Philamore says her team is talking to companies and environmental organizations about using its soft robots to decontaminate polluted waterways and monitor industrial infrastructure.

This doesn’t mean the end of hard-shelled robots.

Part of the problem, says Mark Freudenberg, executive technology director at frog, a design company, is that soft materials break easily, noting that most animatronic dolls like Teddy Ruxpin and Furby have rigid motors and plastic casings beneath their fur exteriors.

To be sure, the nascent soft robot industry lacks an ecosystem of software, hardware components and standards – and some companies have already failed. Empire Robotics, one of the first soft robot gripper companies, closed last year.

RightHand’s Jentoft says the problem is that customers don’t just want a robot, but the whole package, including computer vision and machine learning. “It’s hard to be a standalone gripper company,” he says.

And even if soft robots find a niche, chances are they still won’t replace all the jobs done by human or hard-shelled robots.

Wrench, whose Cambridge Consultants has built its own fruit picking robot, says he expects to see soft robots working with humans to harvest fruit like apples and pears which are harder to damage.

Once the robot has passed through, human pickers would follow to grab fruit hidden behind leaves and in hard-to-reach spots.

“It’s a constant race to the bottom, so there’s a pressing business need,” Wrench said.

Reporting by Jeremy Wagstaff; Editing by Ian Geoghegan