Design and simulation of a soft pneumatic robot with peristaltic locomotion

This project is the starting point of the lab's soft-robot line. It proposes a pneumatic robot that advances inside ducts with the mechanism of an earthworm, and does so with a single air line: the inflation sequence is not imposed by a bank of valves but by the physics of several interconnected elastic cavities. The work develops the model of that behavior, characterizes the candidate materials and lays out a design methodology.

Context

An earthworm moves by alternating the contraction of circular and longitudinal muscles in successive segments, producing a wave that anchors some parts of the body while others advance. Reproducing that wave in a soft robot normally requires one actuator and one valve per segment.

The alternative explored here rests on a well-known balloon phenomenon. The pressure-radius curve of an elastic membrane is not monotonic: it rises to a maximum and then falls. Two connected balloons do not inflate equally; one grows at the expense of the other. In a chain of cavities, that effect can be exploited to obtain sequential inflation.

Conceptual design

Candidate geometries were first explored in VoxCAD, a low-cost voxel simulator that allows configurations to be tried before building a detailed physical model.

Locomotion model

Each cavity was modeled as a balloon with its pressure-radius curve, connected to the others through orifices with a discharge law. Inside the duct, the balloon stops being a sphere and becomes a cylinder with hemispherical ends, which modifies its curve. The model was implemented in Simulink and yields volume, flow, radius and pressure of each cavity over one cycle.

According to the model, the robot would advance 124% of its length per cycle.

For open spaces, a second locomotion mode by rolling was also proposed, inspired by certain caterpillars, in which inflating peripheral balloons shifts the center of mass of a module and makes it rotate.

Materials

Natural latex and a commercial silicone were tensile-tested, and hyperelastic Mooney-Rivlin models were fitted to those data.

With those parameters, the state of greatest deformation of one cavity was simulated by finite elements.

Latex withstands the pressures and deformations of the cycle. The chosen silicone does not: it fails before reaching the deformation that locomotion requires. The work recommends a softer silicone, Ecoflex 00-30, which is the one adopted in the following projects.

Manufacturing

A first approach to fabrication was made with printed molds and inserts for the cavities.

Design methodology

In 2014 there was no established procedure for designing a soft robot. The work proposes one: choose the locomotion pattern, explore it with cheap simulations, build a simplified physical model, select materials and verify their integrity, and only then define manufacturing and control.

What is missing

The main result is a model, and it was not validated experimentally. The 124% advance per cycle is not a measurement. The pressure-radius curves were built from literature parameters and from the uniaxial tests, without an inflation characterization of the actual cavities. Later work showed that this difference matters: the manufactured robot does not deflate in the order the model predicts.

How it fits in Robiolab

This is the origin of a series of three projects: design and model here, then manufacturing and then control. It is also an early example of an idea that is central to the lab: moving part of the control into the robot's body, so that the motion sequence is determined by the mechanics of the material and not by the electronics.