Robot swarm aggregation using an improved Beeclust method

Abstract

Swarm robotics is a topic within multi-robot systems that aims to solve engineering problems by drawing inspiration from the behavior observed in nature by social animals such as ants, bees, fish school, among others. The main challenge in these systems is the design of the controller, which must operate at the level of the individual robot to achieve results at the level of the entire swarm. In previous works, various basic behaviors of social insects have been studied and classified, which have been adapted to the field of robotics to emulate and use in the implementation of controllers and problem-solving. Examples of these behaviors include aggregation, dispersion, and resource searching (foraging). In this project, a simulation implemented in Matlab is presented, which implements a modified version of the Beeclust algorithm, which emulates the aggregation behavior of newly born bees. The modifications made focus on two basic actions of the algorithm: linear velocity and rotation angle. These adjustments have resulted in a 30% improvement in the average aggregation time compared to the original algorithm.