Jun 1, 2017

Robots show a softer side

Image courtesy of Lori K Sanders, Ryan Truby, and Michael Wehner

Advances in materials science and 3D printing have launched a soft robot research boom, potentially easing the integration of robots into our lives. "We've been promised robots among us but the big shortfall has been how they interact with humans," says Harvard researcher Michael Wehner, who built the first entirely soft autonomous robot – called octobot. Three recent examples:

  1. Last week, researchers demonstrated a 3D printed robot that has both soft and hard parts in its legs so it can climb over sand and uneven ground.
  2. Another team recently created an artificial Venus fly trap that is pliable and snaps.
  3. Scientists built a self-powered squishy robotic fish that can tread water and operate for three hours without recharging.

Why it's needed: If robots are going to live and work with humans, they'll need to be safe and perform better around people (more soft on the outside, less hard edges, and more flexible and adaptable). Other applications — like rescuing people after earthquakes or remediation after a toxic chemical spill – require agile robots with the ability to squeeze into tight spaces or cheap, biodegradable ones. And overall, robots need to be able to better deal with the unpredictable world they may occupy, a challenge soft robots might meet.

The impetus for creating soft robots depends on who you ask but the field is emerging in full force from a confluence of the need for robots that can interact with humans, advances in materials science and 3D printing that make it cheaper and faster to print prototypes, and as an answer to engineering problems.

"There was a realization that we haven't fully embraced the materials and phenomena that make animals incredibly good at moving around the world," says Tufts University biologist-turned-roboticist Barry Trimmer. "The moment you try to make a machine move in an unpredictable or changing environment, a lot of the techniques we use in engineering don't work so well."

The challenge – and the solution: Soft materials allow robots to bend, climb, stretch, deform and move their parts in many directions. That also makes it difficult to control them. Mathematically modeling those almost limitless directions and positions in every known, planned or unplanned situation is time-consuming and may not even yield a complete picture of what is happening, says Wehner. Instead, researchers can now print dozens of prototypes in an hour, each a bit different, and then test them. "You can build in an hour or analyze in six months," says Wehner.

A different approach: Imagine a robot picking up a cup. The traditional engineering approach would have the robot figure out exactly where the cup was, where to put each finger of the gripper, and how much force to apply. It would calculate each position and all the torques to move the cup to a particular place. But humans aren't even that careful about how we control our limbs. "We're very precise and accurate but if you look down at the details, there is a sloppiness going on but the sloppiness is fine. It doesn't matter. It's a different strategy," says Trimmer. Our mechanics compensate for that sloppiness and allow the brain to offload the fine calculations of movement to the limbs themselves. "The complexity of the mechanics and materials is actually being exploited. It is the solution to problem," he says. "It's the way animals work but it's not the way we make our robots now."

Key advance: Flexible sensors – a burgeoning field of research in and of itself that includes liquid metal versions, ionic types that rely on a salt solution that changes resistivity with shape and optical sensors - are all being developed to allow robots to grip, move and embrace uncertainty in their environments like animals. Wehner and his colleagues are looking to embed these sensors in and on their robot so it can better process itself and the outside world. "Soft robots require us to fundamentally rethink sensors," he says.

Big hurdle: Soft robots need soft motors and that, says Trimmer, is what is limiting the field currently. There are polymers that change size or shape when a current passes through them but they aren't widely used. Other researchers are using pneumatic motors. Trimmer is trying to engineer muscle tissue and use it as the motor but he is far from getting the right shape and size. His vision: "Robots that - like humans – are powered by cheap hydrocarbons (sugars, fats). Then it can live in the same house as you because it isn't producing gasoline fumes or using batteries."

Far out: Trimmer hopes to build robots from living tissue or biological materials like silk, which can be soft or stiff, is compatible with the human body, and can be biodegradable. "There is no reason on the planet why you couldn't make soft robots out of silk."

Go deeper