Communications of the ACM
Robots Learn to Share, Validating Hamilton's Rule
Ecole Polytechnique Federale de Lausanne professor Dario Floreano
Photo courtesy of Dario Floreano
Researchers at Ecole Polytechnique Federale de Lausanne (EPFL) and the University of Lausanne are developing robots with altruistic traits based on Hamilton's rule, which states that an organism is more likely to express altruism with others who possess many genetic similarities.
"We have shown that Hamilton's kin selection theory always accurately predicts the relationship between the evolution of altruism and the relatedness of individuals in a species," says EPFL's Markus Waibel.
The robots start with a random series of coded information to complete a task. Only those robots that successfully complete the task are able to pass on their code to the next generation. After hundreds of generations, the robots become more efficient and learn to work in groups.
The researchers created groups of relatedness that are the equivalent of clones, siblings, cousins, and non-relatives. The groups that shared traits correlating with Hamilton's rule completed the tasks better and passed their genes on to the next generation.
"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," says EPFL professor Dario Floreano.
From Ecole Polytechnique Federale de Lausanne
View Full Article
No entries found