Computing and Informatics

It is considered a difficult task to design a controller from scratch for a robot in a dynamic environment. Evolutionary and genetic algorithms are frequently used to find a solution with desired properties. The evolution is supposed to run on one robot or agent processor. In this article we explore the possibility of dividing the genome among several robots. Robots exchange among each other successful controllers in the form of chromosomes that code for the weight vector in a neural controller. Thus; the evolution can be faster because it runs in a parallel manner over the whole robotic society, not in one robot. Second interesting point is the exchange of experience among robots. Robots can send each other parts of the controllers that they evolved on their own. They can learn behavioral strategies that other robots developed. We describe experiments done with small Khepera robots in the domain of a benchmark test for evolutionary robotics. We compare the behavior generated by one robot based on its individual evolution with the robot that profits from sharing the full set of chromosomes with the other robot.