Evolving 3D shapes: StFX student building a system using evolutionary computation 

Student sitting at desk with computer
Ethan Heavey

For the last number of months, Ethan Heavey of Fernie, BC has been using Artificial Intelligence and Machine Learning to create arbitrary 3D objects with a 3D printer as part of his honours thesis.  

Mr. Heavey’s supervisor, computer science professor Dr. James Hughes, says what is particularly interesting is that Mr. Heavey is using the AI in an unusual way—the AI has no real objective it is trying to achieve, which is very atypical. 

“I'm trying to build a system that uses evolutionary computation to evolve 3D shapes by taking input from a user,” says Mr. Heavey who graduated from StFX in May with an advanced major in computer science and a major in business administration. He will start a masters in computer science this fall, studying under Dr. Hughes and Dr. Milton King. 

He will also continue to work on the project this summer to polish the work for publication. 

“By selecting a shape that appeals most to them, the user will be presented with different generations of shapes, taking them on a unique evolutionary journey into all the possible shapes that can be created using the system,” he says.  

Mr. Heavey says the idea for the work originated while he was reading "The myth of the objective," a novel Dr. Hughes loaned him during his third year of study. In the novel, Picbreeder, a similar system that evolves 2D images, is described and elaborated upon. “I wanted to expand Picbreeder by a dimension and test if the conclusions reached by the co-authors apply in three dimensions.”

He says the research has primarily focused on how utilizing evolutionary computation without a specific objective for the system to strive towards can lead towards. Mr. Heavey is now working to transfer the system to a website so anyone can access it whenever they like. 

“Increasing the number of users and the number of shapes said users create helps us to identify unique paths between shapes.”
As for anything surprising that came from the research?

“Definitely!” he says. “Even though the system had no clear goal other than ‘evolve,’ it was able to create complex shapes. However, the path User 1 had taken to create Shape A may never be recreated by User 2 in an attempt to recreate Shape A. Even though we know the shape exists and that the system can evolve it, there is no guarantee the same shape can be created twice through the same means. Each shape requires a unique journey, which can provide some interesting insight into evolution and natural selection with living organisms.”

This research is, in part, made possible by the Government of Canada Research Support Fund.