On the evolution of self-organised role-allocation and role-switching behaviour in swarm robotics: a case study

Abstract

In spite of its significance for the adaptability of autonomous robotic swarms, the dynamic allocation and re-distribution of robots to tasks (i.e., role-allocation and role-switching be-haviour) is still a design challenge in swarm robotics. This study investigates a simulated foraging scenario in which the variability of the environmental conditions requires that robots switch between two roles (i.e., foraging and nest-patrolling). To the best of our knowledge, this is the first simulation study that demonstrates that role-allocation and role-switching behaviour can be evolved using dynamic neu-ral network controllers for robots with minimal communica-tion capabilities. Initial analyses of the best evolved teams shed light on some of the characteristics and robustness of the strategies used by these teams to repeatedly face this task.