The EPFL (Ecole Polytechnique Federale de Lausanne) will definitely not stop to talk to her, in robotics. A team of Swiss researchers consisting of robotics Dario Floreano, Sara Mitri, Stéphane Magnenat and Laurent Keller of the biologist, has developed a simulation composed of virtual robots that can understand what factors determine how communication arises in the evolution of social organizations. The virtual robots are used compounds of evolutionary algorithms, which enable them to respond and adapt to their environment. Communication is critical to the ecological success of social organizations. But the study of the evolution of communication is made extremely difficult for the absence of trace fossils of communication as it existed in the primitive social animals.
To perform their study, the researchers then used the simulation (the simulator is Enki, a 2D simulator in C + +). They studied the behavior of hundreds of colonies composed of 10 individuals. The development of these settlements has continued for 500 generations. Once the robot “virtual” presented interesting features, the program was transferred to real robots, this time.
Presentation of experiences
The robots are placed in an unfamiliar environment, provided a reserve of food and a pool of poison. Each robot is able to move and differentiate the poison food. Each robot has basically a simple program, encoding settings variables in a “genome”, a sort of database that can change over generations. The parameters in question encode for example the importance of altruism, or the spirit of competition. When the “reproduction”, the most powerful individuals will be selected, and will produce a new generation by passing each genome, which in theory combine the benefits of both parents.
According to Laurent Keller, “under certain conditions, a sophisticated communication has developed. We have seen colonies use lights to indicate they had found food and others reported the poison.” The use of primitive codes could well prove to be the root of language.