The impact of population composition for cooperation emergence in evolutionary robotics

Abstract

Communication is an important tool for evolutionary robotics. Some important aspects are the emergence of signals, the environment, and manipulation of social and evolutionary variables. In this paper we focus on social aspects related to exploration in poisoned and food environments. These aspects are as follows: a) intermediate levels of heterogeneity in population of evolutionary robots, and b) cooperation of robots for fitness contribution to regulate the emergence of communication signals. The FARSA simulator and Marxbot robot are used in order to optimize the weights of neural networks using a steady state genetic algorithm. A basic communication system is developed based on color LEDs and linear cameras.