Evolutionary robotics is a new fascinating field of study which leverages Darwinian principles to develop new robots.
In a new study published in Frontiers in Robotics and AI, scientists added more intricacy to the field by showing the very first time that robot evolution is affect by developmental factors, just like in biological evolution.
In evolutionary robotics, a sort of "gene pool" is produced, which generates genomes that encode the various control methods of a robot. Each robot is then permitted to act and carry out tasks based on its "genetics", and the robot's fitness is rated based on how well it does a particular job. The robots are then "bred" by trading genetic material with each other; similar to genetic material is recombined in sexual reproduction.
In living organisms, the genome is impacted by events that take place during their lifetime, also known as ‘epigenetic’ modifications. In biology, this interaction between evolution and development is referred to as “evo-devo”, which has highlighted the significance of non-genetic variables on an organism's phenotype.
In their report, the study team said the evo-devo challenge for robotics researchers is to create physically embodied systems that incorporate the “three scales of time and the processes inherent in each: behavior, development, and evolution.”
“Because of the complexity of building and evolving physical robots, this is a daunting challenge in the quest for the evolution of things,” the study authors wrote. “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."
In their study, the researcher set out to generate a system that might be used to examine how evolutionary and developmental variables work together in robot evolution, and how developmental variables impact robot evolution. While previous research studies have centered on the effects of evolution in physical robots, this is the first time that scientists have also factored in the epigenetic aspect in this kind of experiment.
"An explicit evo-devo approach has proven invaluable in the evolution of artificial neural networks. Development serves as a new type of evolutionary driver – alongside the genetic factors of mutation, recombination, and selection," the study said. "We note that what is missing from evolutionary robotics is not development per se but rather physically embodied development. We take the first simple steps toward combining the two by examining the interactions of epigenetic and genetic factors in the evolution of physically embodied and simulated robots."
The study team assessed the fitness of individual robots based on how they performed in two tasks: light gathering and obstacle avoidance. A randomized mating algorithm was used to find out which parental "genomes" should be blended to create the next generation of robots. The researchers repeated their process until 10 generations of robots had been created and ranked. In addition to the experiment on physical robots, the team also produced and evolved modeled robots so they could compare the results of both the physical and modeled robot populations.
The outcomes indicated that robot populations with an epigenetic factor progressed in a different ways than populations where development was not factored in. While robots did not evolve greater light capturing abilities, the team said they were excited about the outcomes, since purpose of this preliminary research was to show the significance of including epigenetic factors in robot evolution and to create a strategy that makes this study possible.
"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,” the study team wrote. “The results do indeed suggest demonstrate the broad importance of including EOs in investigations of evolvability.”
Image Credit: shutterstock.com/lindabucklin shutterstock.com/kit8.net