2017 Academy Award nominations: The snubs, surprises and what happened to Scorsese?

Watch it jump! 3D-printed hopper could lead to better rescue robots

If you think robots involve metal bodies and squeaky hinges, think again.

Engineers have designed and built a frog-like jumping robot that incorporates hard and soft parts — and they’ve done it with a 3D printer.

Powered by a mix of butane and oxygen, these adorable hopping bots can jump two and a half feet high and half a foot sideways, and can survive more than 30 jumps without breaking. The machines, described in the journal Science, could help engineers design more robust robots, ones that could be used in future rescue missions.

"You could use it, say, for exploring unknown or unstructured spaces where it’s not enough to have wheels or roll around. You might need it to hop or jump to overcome obstacles," said Carmel Majidi, a mechanical engineer at Carnegie Mellon University, who was not involved in the study.

If you watched the DARPA Robotics Challenge last month, you may have noticed that the robotic competitors, like your prototypical bot, are generally made out of hard metallic and plastic parts.

That’s well and good for many purposes, but they run the risk of accidentally causing damage or injury. Soft parts would be a safer option, and they’re also more useful for tasks that involve, say, picking up a small or delicate object. Try to shake hands with a robot in the future, and you'll learn (painfully) the true meaning of a steely grip.

But a completely soft robot has its drawbacks too, said Michael Tolley, a mechanical engineer at UC San Diego who co-led the study. Tolley, who recently helped create an origami-inspired self-assembling robot, says he worked previously on completely soft robots, but the pneumatic machines worked at a very slow pace.

So the scientists decided to build a faster robot with hard and soft elements — and they chose to make a jumping robot. This little bot would have to move quickly to make those leaps, but it also would have to be tough enough to survive the landing.

“Jumping seems like a challenging problem … that really, in my mind, showcases some of the benefits of a soft system,” Tolley said.

Here’s the problem: Soft and hard parts often don’t work well together in these hybrid systems. That boundary between the two is an easy breaking point, the roboticist said.

“It’s sort of well-known in material science that if you have a very rigid thing connected to a very soft thing, you get stress concentrations at that interface, and that can lead to all sorts of problems,” Tolley said.

But biology deals with this problem all the time, he pointed out: Living creatures typically have hard parts along with soft ones.

“So we looked to nature," Tolley said. The octopus, for example, has a sharp, hard beak, but the material in it gradually transitions along its length from the hard beak to the soft body. 

"So there’s no abrupt transition, which sort of makes sense intuitively," he added. "Everywhere we look in nature, you see what we call 'material gradients.' ”

Layering several materials — each one slightly softer than the previous layer — appeared to be the solution. So the researchers used a 3D printer that creates objects using two different materials (one hard and one soft).

They built the top half of the rounded robot using both the hard and soft substances, but in different ratios. The first layer was almost entirely hard material with a little soft material, the next layer had slightly more soft stuff in it, and so on until the ninth, softest layer.

To compare how well this material gradient worked, the scientists also made two other robots with different tops: a hard, rigid top and a mostly soft top (a small portion in the robot's core had to remain hard because that’s where the batteries and controller were housed).

When the scientists tested their jumping ability, they found that the robots with the soft tops were pretty pathetic jumpers because, without much hard stuff in their bodies, they couldn’t effectively push against the ground. The hard-topped bot was the best jumper of the three — but its rigid body quickly suffered cracks upon landing.

The robot with the layered top, gradually shifted from hard to soft, was the clear winner — it jumped high enough to be effective but its top was far better able to withstand the landing impacts.

In fact, the layered robot was better at withstanding the landing stress than both the hard and soft versions, the study authors wrote.

“We find that the rigid and flexible top robots only absorb 13% and 73% (respectively) of the impact energy that the gradient top robot absorbs,” they wrote.

With their current 3D printer, the scientists’ design was limited to a maximum of nine layers, using only the two given materials to print each layer. Ideally, you’d want more layers to make the transition even smoother, Tolley said — but those developments are a matter of time.

“I imagine in the future more combinations will be possible and more materials in general will be possible,” Tolley said.

Previous work on these kinds of soft robots often had to use injection molds and other traditional manufacturing methods to make the parts — which can be costly and inefficient. But the researchers showed that 3D printing could make it much easier to create and tune these material gradients and build more flexible-but-robust robots, he said.

“Now we have this freedom to fully design and determine how these things are arranged spatially, and that’s what I think is really exciting,” he said.

Future robots using these soft parts would be much safer for humans to handle — and they could be useful in rescue operations, as they can leap over obstacles that might stop a wheeled robot in its tracks.

And when properly integrated, soft parts don't just make robots safer to work with, researchers said, they can make the robots more robust.

Ideally, "you need a robot that is soft enough that it can squeeze through tight spaces and is also robust and elastic so if it does hit into something ... you have soft materials to distribute those impact forces so it doesn’t cause damage to the robot," Majidi said.

Love robots? Follow @aminawrite for more weird science news.

Copyright © 2017, Los Angeles Times


July 10 1:23 p.m.: This article has been updated with additional details as well as with comment from Carmel Majidi, a Carnegie Mellon University mechanical engineer who was not involved in the study. 

This story was originally posted at 2:59 p.m. July 9.