Evolutionary utility of emerging communication systems and Singnal Complexity in Robotics

Abstract

Communication allows robots to exchange information and coordinate to solve complex tasks. Evolutionary methods are used to emerge communication in groups of robots. However, many variables are involved, and they must be studied to understand their influence on the emergence of communication systems. Thus, a study is presented to try to understand if a standard approach is more effective than using Evolutionary Robotics in order to produce development signals. For this, an environment is adapted to solve the task of discrimination between food and poison areas, which favors the emergence of signals. The experiments are performed in a simulated environment using FARSA and the Marxbot platform. Furthermore, we set three experiments varying the complexity of the emerging communication system in order to measure fitness levels. Finally, we found that a communication system involving more signals significantly increases the task’s fitness.