After each robot completed its tasks, pairs of individual bots were sorted into five ranks by order of fitness. From there, a randomized mating algorithm was used to determine which parental genomes would be combined to produce the next generation of robots.
In this case, the genomes consisted of binary code that allowed for different possible wiring of the bot's hardware setup. The emerging phenotype — the physical expression of the gene — was modified in each generation by altering their wiring in accordance with the new genetic information. The process was repeated until 10 generations of robots had been created and ranked by fitness.
RELATED: Killer Machines and Sex Robots: Unraveling the Ethics of AI
The researchers threw in another twist as well, based on a particular aspect of evolutionary theory. In living organisms, genomes are affected by development as well as evolution. In this context, development refers to events during a single lifetime that lead to epigenetic changes. This interplay between evolution and development is sometimes referred to as evo-devo, and it represents a discrete field of study in evolutionary developmental biology.
It gets complicated, but the upshot is that the Vassar experiment was the first to introduce developmental variations in an experiment with physical robots, according to the researchers. The core idea was to study how genetic (evolutionary) and epigenetic (developmental) factors interact in robotic evolution. Similar studies have been applied in the field of artificial intelligence and neural networks, but the Vassar team was interested in the potential future of physically embodied robots.
"For roboticists, the evo-devo challenge is to create physically embodied systems that incorporate the three scales of time and the processes inherent in each: behavior, development, and evolution,” wrote project leads Jake Brawer and Aaron Hill, who authored the report with four other colleagues. “Because of the complexity of building and evolving physical robots, this is a daunting challenge in the quest for the 'evolution of things.' As an initial step toward this goal, in this paper we create a physically embodied system that allows us to examine systematically how developmental and evolutionary processes interact.”
RELATED: Stopping Killer Robots at the Source (Code)
It turns out that the experiment didn't reveal anything particularly dramatic. The robots didn’t evolve better light-capturing or object-avoidance skills. But the experiment did reveal the importance of tracking the developmental factor in evolutionary robotics.
"It is important to note that our goal was not to show adaptive evolution per se, but rather to test the hypothesis that epigenetic factors can alter the evolutionary dynamics of a population of physically embodied robots,” wrote Brawer and Hill.
Notably, all the bots had lost mobility entirely by the end of the experiment, since the mating algorithm allowed low-fitness individuals to remain in the gene pool and reproduce. So maybe there’s still hope for us after all.
WATCH: Heart Cells Are Bringing Robots to Life