Robots with the ability to evolve quickly show that robots within the same family tend to share more with closer relatives than with distant relatives or strangers
Researchers from École Polytechnique Fédérale de Lausanne (EPFL) and the Université de Lausanne in
Dario Floreano, study leader and EPFL robotics professor, and Laurent Keller, a biologist from the Université de Lausanne, have joined forces to study how altruism came about in social creatures that were born to understand "survival of the fittest."
Altruism is the selfless concern for the welfare of others. It is demonstrated in animals like worker ants, which are sterile and give up the transmission of their own genes in order to guarantee the survival of the queen's genetic makeup. This type of altruism is known as kin selection, where the individual makes personal sacrifices in order to insure the survival of a relative's genetic code.
In 1964, a biologist named W.D. Hamilton proposed that if an individual member shares food with its family, it decreases its own survival yet increases the chance of survival for family members that can pass along family genes. Hamilton's rule of kin selection states that an organism is more likely to share its food with another that is genetically close rather than those who are not.
Floreano and Keller began experimenting with this theory using robots. In previous studies, the robots would perform simple tasks like pushing seed-like objects across the floor to specific destinations. The robots would evolve over several generations, and those that could not push the objects to the correct destination were pulled from the study, unable to pass their code along. On the other hand, robots that were able to perform the desired task were allowed to pass their code on, and this code was mutated and recombined with other robots in the next generation.
Now, Floreano and Keller are testing the evolution of altruism through the use of robots that evolve quickly. The robots are able to evolve through the use of simulated gene and genome functions. Some robots are created as clones, siblings, cousins, and complete strangers. The difference between this new study and the older ones is that robots now have the option to share a seed once it is pushed to the desired destination. Over the course of 500 generations of experiments, the researchers observed when robots would share, when they wouldn't, and what the consequences were for both options.
The results of the study showed that Hamilton's rule of kin selection was exactly right. Groups of robots that were related shared the seed, and the family became strong and was able to pass its code to the next generation. Hamilton's original theory takes into account both a limited and isolated vision of gene interaction while these gene simulations in robots "integrate effects of one gene in multiple other genes," and Hamilton's theory was still correct.
"We have been able to take this experiment and extract an algorithm that we can use to evolve cooperation in any type of robot," said Floreano. "We are using this altruism algorithm to improve the control system of our flying robots and we see that it allows them to effectively collaborate and fly in swarm formation more successfully."
This study was published in PLoS Biology.
No comments:
Post a Comment