John Rieffel

Assistant Professor
Computer Science Department
Union College


On this page:


Soft Robotics

Evolutionary Fabrication

Introduction: Mechanisms as Minds

My research explores how mechanisms can act as minds. Specifically, I use genetic algorithms (GAs) to understand the ways in which natural and artificial systems can leverage complex real-world dynamics to their advantage. In the realm of robotics, I am exploring how aspects of control which are normally attributed to cognition (in animals) or computation (in robots) can instead be ``outsourced'' directly into body dynamics. In the realm of 3-D printing, I am exploring how GAs running on rapid prototyping 3D printers can be used to invent entirely new physical objects and entirely new forms of fabrication.

What distinguishes this work from other GA-based robotics research is that it is by necessity physically grounded, eschewing simulation. Highly collaborative and interdisciplinary, the insights which result from this research have applications ranging from the creation of search-and-rescue robots to the automated invention of patentable designs.

Soft and Amorphous Robotics

Tensegrities, like this toy here, are composed of rigid rods and tensile strings, and maintain their shape due do a complex interplay of tension and compression.

Tensegrities are structures composed only of rigid compressive elements and tensile string elements. They hold their shape due to a careful balance of opposing tensile and compressive forces. This "pre-stress stability" imposes complicated dynamics like vibration, and therefore makes them challenging to control through conventional means.

Our research involves exploiting (rather than attenuating) these dynamical phenomena in order to produce locomotion.

My students have a blog describing their progress

We do this by mounting pager motors on the rods and using a genetic algorithm to find motor frequencies which induce locomotion. A digital video camera records motion and measures fitness.

Evolutionary Fabrication

Evolutionary Fabrication (EvoFab) seeks to combine recent innovations in desktop rapid prototyping with the creative power of genetic algorithm-based Evolutionary Design in order to create a machine which can automatically design and build completely novel objects.

The EvoFab(0.2) consists of a Fab@Home prototyper which prints objects out of play-doh, a camera to measure the fitness of printed objects, and an arduino-controlled LEGO conveyor belt to move objects once they are printed.

My students have a blog describing their progress. Please see my Videos page for videos of these experiments.