Optimization of felf-organized flocking of a robot swarm via evolutionary strategies

This work presents the evolutionary optimization of a self-organized flocking controller for a swarm of mobile robots. The robotic platform utilized is Kobot, a robot developed specifically for swarm robotic studies. In previous work, a successful self-organized flocking algorithm was proposed, which depends on two basic behaviors of heading alignment and proximal control against nearby robots and obstacles. The controller is subject to many parameters, which include the relative weights of the two behaviors, and the maximum speed, the proportionality constant, and the maximum angular velocity of an individual robot. The possible dependence of the parameters on each other render a grid optimization infeasible. In this work, we present the partial results of an optimization of these parameters via evolutionary strategies by using a set of explicit metrics.