Squishy, Soft Robots Crawl Their Way to the Cutting Edge of Science

A new breed of robots based on spineless creatures such as starfish and caterpillars could change the way humans interact with machines.

  • Share
  • Read Later
Rob Shepherd

It looks like footage taken from a deep-sea submersible. A small creature fills and deflates sacs to crawl across the ground, not a bone present in its translucent body.

It’s not organic; in fact, it’s a robot designed by Dr. George Whitesides and his team at Harvard. In the future, soft robots like it will do everything from search disaster sites to inspect our internal organs to assist the elderly.

“The field is so young,” says Whitesides. “As we see it, we’re in the stage where almost everything we try works.”

While the idea of biomimetic robots—biologically inspired robots that mimic animals such as dogs and ostriches—isn’t new, soft robotics is in its infancy, really only taking off within the last five or six years.

(PHOTOS: Cinema’s Most Memorable Robots)

It’s a field that throws a lot of the conventions of robotics out the window. Biologists, computer scientists and chemists work side-by-side with mechanical engineers, trying to create brand-new models for unprecedented problems.

Instead of metal rods or sheets, most soft robots utilize materials similar to the elastic polymers used to make Dr. Whitesides’ “starfish,” which have the added advantage of being quickly and cheaply produced by another technology that’s also been on the rise: 3D printing.

“In experimentation, you have an idea, you try it, it fails, and then you look at the failure and you try to find something that works better,” says Whitesides. “You want to do that rapidly, to get your mistakes out of the way.”

The 3D printer in Whitesides’ lab allows for that trial-and-error process to be sped up. The result: A functional crawling and undulating robot aimed at helping first responders search through rubble for disaster victims.

He’s also working on a robot meant to replace hard clamps in operating rooms, gently moving organs aside so that surgeons can perform delicate procedures without worrying about damaging the body.

While small successes come quick in new fields like this, there are some serious challenges to conquer before we start seeing soft robots crawling around outside of university laboratories.

article continues on next page…

“The two things that really have to be addressed are actuation—how do you actually make these things move—and the other thing is creating a new theory for controlling soft materials,” says Dr. Barry Trimmer of Tufts University’s Soft Robotics Research Group. “The mathematics are extremely complex. It’s a very complicated problem that hasn’t been addressed by enough people or with enough money.”

The computer models for controlling conventional robots are pretty well-established; two rigid surfaces, connected by a joint, can only relate to each other through a certain number of angles.

In soft robotics, however, it’s a whole different ballgame. One surface might be stretching outwards while another is constricting, meaning that relating any two points to each other can become a real pain.

That would explain the draw for computer scientists, who are intrigued by creating new, wildly innovative computer models. Neurobiologists like Dr. Trimmer come for another reason.

“I’m already studying what I like to call a ‘living prototype’ for the robots,” said Trimmer, who is developing a robot that inches forward like a caterpillar. “Soft robots should be designed to operate in natural environments and interact with humans, and most robots just don’t do that.”

One way to create more nature-friendly robots would be to make them out of biopolymers that would decompose in the wild and be safe to use in the human body, which could eventually lead to robotic endoscopes that could provide doctors with previously impossible footage.

Dr. Trimmer is also working on a machine called the GoQBot which aims to solve a fundamental problem in soft robotics: Moving soft materials with compressed air or hydraulics is inherently a slow process. GoQBot attempts to remedy this by curling into a ball and rolling, just like real caterpillars do.

Another robot in the very early development stages is a crawling water bottle which would find victims stuck in rubble, set off a beacon and then provide them with around a liter of water, which could keep them alive for an additional day while search and rescue teams locate them.

A plus when it comes to basing small robots off of simple creatures such as caterpillars and starfish is the low cost; each is developed to be disposable after its task is done.

Of course, not everything in soft robotics resembles an insect or sea creature.

article continues on next page…

At Worcester Polytechnic Institute, Eduardo Torres-Jara is working with tactile sensors and soft materials to create a robotic arm that can gently manipulate objects and be interacted with safely by humans.

His goal is to create a robot with a soft exterior and an abundance of sensors, the idea being that when you grab it gently on the arm and pull it one direction, it would sense and then follow you. In the end, it would have to be non-threatening enough to make elderly or disabled people feel safe, with the capability of adapting to situations and taking commands through physical interaction, as opposed to through a computer network and visual or infrared sensors like most robots do.

“Take how we interacted with computers,” says Torres-Jara. “Before it was just people making holes in cards, then it moved to a keyboard and a screen, and now you have things like the iPhone and the iPad, which are very easy and intuitive to interact with. That’s also going to happen with robots. Instead of people sitting at a computer and programming them, the robots themselves will be able to get information directly from the people who are trying to train them.”

Ultimately, the goal for almost everyone in soft robotics is the same. Approach the functionality and efficiency of nature’s perfect actuators: muscles.

“The muscle is an incredibly sophisticated biological structure, and understanding how to mimic its structure and its function is just not something we know how to do,” says Whitesides. “The ability to come up with something that works like a muscle, a real muscle, with the same functions would be an enabling technology that would allow us to do all kinds of things.”

Dr. Trimmer over at Tufts sees a different, more radical path for soft robotics. “I was scratching my head and thought to myself, ‘Why are we using artificial muscles? Why don’t we just use muscles?’”

His idea is to eventually grow muscle cells and control them with an implanted chip. While the idea of a robot made of flesh might be unsettling for some, Trimmer claims the lack of a nervous or reproductive system makes it no different from, say, the meat being grown in a lab by scientists in the Netherlands.

No matter what the field of soft robotics looks like in 10 or 20 years, one thing’s for sure: The days of thinking of robots solely as stiff, mechanical creations are over forever.

MORE: The Next Step in Robotic Neurosurgery

  1. Previous
  2. 1
  3. 2
  4. 3