Evolutionary Swarm Robotics: Genetic Diversity, Task-allocation and Task-Switching

Abstract

The goal of this study is to investigate the role of genetic diversity for engineering more resilient evolutionary swarm robotic systems. The resilience of the swarm is evaluated with respect to the capability of the system to re-distribute agents to tasks in response to changes in operating conditions. We compare the performances of two evolutionary approaches: the clonal approach in which the teams are genetically homogeneous, and the aclonal approach in which the teams are genetically heterogeneous. We show that the aclonal approach outperforms the clonal approach for the design of robot teams engaged in two taskallocation scenarios, and that heterogeneous teams tend to rely on less plastic strategies. The significance of this study for evolutionary swarm robotics is discussed and directions for future work are indicated.