Selection pressure shapes epistatic gene interactions in the evolution of robots

This study reveals how gene interactions influence the traits and behaviors of robots designed using bio-inspired algorithms. Through the application of measurement techniques from experimental biology to artificial gene regulatory networks, it is shown that epistasis--a non-additive effect of gene interaction--can be shaped by evolutionary processes in artificial systems, while influencing the quality of the robot traits. These insights offer a path toward improving the explainability, reliability and safety of robotic systems. 

[Karine Miras and Anton Feenstra]

pre-print