This January, a group of researchers in Cornell’s Collective Embodied Intelligence Lab overcame a barrier in soft robotics by designing and modeling a system that takes advantage of fluid viscosity.
Soft robotics is a subset of robotics that utilizes soft materials to create systems capable of moving independently and mimicking natural organisms. Rather than using hard materials like metal or plastic, soft robots use rubbery materials that resemble muscles or skin. They extort the qualities of a material that predispose it towards a particular motion — such as how the traits of a slinky lead it to naturally “walk” down the stairs on its own. The research group, led by Prof. Kirsten Petersen, electrical and computer engineering, used small tubes and a bellow system, which is a grouping of small, inflatable balloons, to create a robotic limb capable of an unprecedented range of movement.
In the recent paper, lead author Yoav Matia, a postdoctoral associate in the Sibley School of Mechanical and Aerospace Engineering, employs predictive modeling — which uses mathematics to describe and forecast physical phenomena — to summarize the experimental results. The outcome is a product of years of work on soft robotics within the Collective Embodied Intelligence Lab.
Soft robotics often uses the flow of fluids through tubing and valves to produce movement. In doing so, viscosity, which is the resistance of a fluid to flow, delays its movement. Viscosity can differ between fluids and the size of the channel they move through.
“You can imagine if you had a really long straw, and you were trying to suck air through, it’s not a problem, but if you’re trying to suck up water or a cold slushy it takes a while to suck it up,” Petersen said.
This delay in movement makes traditional soft robot motion unpredictable, and as a result, researchers often design soft robots with large tubes to avoid high viscosity.
Despite its unpredictability, Petersen’s team utilized viscosity as the driving force behind the robot’s motion. They created a core unit limb capable of movement when air was pushed through the tubes with a syringe. The unit consists of two stacks of balloon-like bellows connected by very thin tubes and a syringe that inflates or deflates the bellows with air. The small size of the tubes induces viscosity.
“It’s sort of like a bendy straw, but the big difference is that in between the bends, instead of having a big channel, we have really really thin channels so that we can use the fact that there’s a delay for the air to propagate through all of the bends,” Petersen said.
This delay in movement causes uneven pressure distribution — the differences in force from the air on the walls of the bellows. As a result, the stacks of bellows contort in different directions depending on the way the air diffuses. For example, if the left of the two stacks was more pressurized, the limb would appear to curve, bulging towards the left.
Pressure can be carefully controlled, creating an infinite range of available movement with only one syringe inputting fluid.
“There’s tons of applications for traditional robots but more and more people are also starting to look at these soft robots,” Petersen said. “ If you’re clever in your way of controlling them, the entire robot becomes much much simpler.”
The use of small tubes to move the bellow system also creates greater flexibility in motion over time. Limb motions develop stroke-like fluidity, with varying speeds throughout the cycle that can be manipulated to produce different movement patterns.
The intricacies of the movement were summarized in the paper by a predictive model developed by Matia. The mathematical model closely examined the physical phenomena behind this motion. Matia built the model from the ground-up, combining intricate physics theory, such as viscous dynamic theory, with the material properties.
The model isolated five variables experimenters could change, which inevitably alter limb movement patterns: bellow configurations, fluid properties, fluid input, viscosity and initial pressure. They can be combined in any way but are powerful alone. For example, changing only the tube alignment can mean the difference between long strides and a trot when the limbs are used as legs.
Identifying these parameters in conjunction with the model gives other researchers in the field a recipe to build their own soft robot.
“We were trying to say: Here’s a whole new type of soft robot which we haven’t seen before, and here’s all of the foundational knowledge you need in order to play with the science in this space,” Petersen said.
As robotics technology continues to advance, soft robotics becomes increasingly appealing due to its embedded safety, simplicity and functionality. The work done by Petersen’s team carves a clear path forward.
“We managed to identify a void in what soft robots had exploited in the past,” Petersen said. “This is one example — there’s so much we can do from here.”
Laine Havens is a member of the Class of 2025 in the College of Arts and Sciences. She is a contributor for the science department and can be reached at [email protected].