Simultaneous Localisation and Optimisation for Swarm Robotics

Abstract

Collective search mechanisms usually assume that the positions of all particles are known. In robotic applications information on the environment, such as the position of the robots, is not known, but needs to be measured. We present the Simultaneous Localisation and optimisation method that combines a localisation scheme based on the decentralised GPS-free Directed Localisation algorithm with Particle Swarm Optimisation to perform a simulated robotic search. Our experiments show that our algorithm is capable of finding a goal in a fitness landscape, that higher measurement errors lead to more exploration and less exploitation and that there is a minimal particle to particle distance below which the algorithm shows no further convergence. We hope that our algorithm can serve as a blueprint that enables the use of swarm intelligence algorithms in more robotic applications than before.