One advantage of unconstrained soft robots is their ability to mechanically adapt to their surroundings and tasks. Now they are ready to become more flexible and in control.
A team of researchers led by Kirstin Petersen, assistant professor of electrical and computer engineering at Cornell University, has designed a new — and surprisingly simple — system of fluid-powered actuators that enables soft robots to achieve more complex movements. The researchers achieved this by taking advantage of the very thing — viscosity — that previously impeded the movement of soft, fluid-driven robots.
The team’s paper, “Harnessing Nonuniform Stress Distributions in Soft Robotic Actuators,” is published in advanced intelligent systems.
Petersen’s Embodied Collective Intelligence Lab has been exploring ways to take a robot’s cognitive abilities and behaviors and unload them from the “brain” into the body, via the robot’s mechanical feedback. By reducing the need for explicit arithmetic, a robot can become simpler, more robust, and less expensive to manufacture.
“Soft robots have a very simple structure, but they can have more flexible functionality than their rigid cousins. They are a kind of ultimately embodied intelligent robot,” Petersen said. “Most soft robots these days are fluid-driven. In the past, most people looked at how they could generate additional profits by embedding functionality in a robot’s material, such as elastomer. Instead, we asked ourselves how we could do more with less by taking advantage of how robots interact. liquid with that substance.”
Petersen’s team connected a series of elastomer bellows to thin tubes. This configuration allows for antagonistic moves – one that pulls and one that pushes. Small tubes stimulate viscosity, causing the pressure to be distributed unevenly, causing the actuator to bend into torsions and different movement patterns. This would normally be a problem, but the team has found a clever way to take advantage of it.
The researchers developed a fully descriptive model that can predict potential movements of an actuator – all with a single fluid input. This results in an actuator that can achieve much more complex movements, but without the multiple inputs and complex feedback control required by previous methods.
To demonstrate the technique, the team made a six-legged soft robot, with two injected pumps on top, that walked at 0.05 body length per second, and also crouched. But this is only the beginning of the possible permutations.
“We’ve detailed a whole range of approaches by which you can design these actuators for future applications,” Petersen said. “For example, when using actuators as legs, we show that once you cross one set of tubes, you can go from an ostrich-like gait, with a really wide stance, to an elephant-like trot.”
The new fluid-powered actuator could be used for different types of devices, such as robot arms, and Petersen is interested in exploring how placing bellows in three-dimensional configurations will lead to more useful motion patterns.
“This is basically a whole new subfield of soft robotics,” she said. “Exploring this space will be so much fun.”