Abstract

In swarm robotics, simulation is an indispensable tool to study algorithms involving large numbers of robots. Despite its importance, the creation of a simulator that is truly accurate, efficient, and capable of supporting a wide set of use cases has remained an unsolved problem for a long time. In this paper, we present ARGoS, the first simulator able to show accuracy, flexibility and efficiency. Accuracy concerns the correspondence between results obtained in simulated and real-world experiments; flexibility allows the simulator to support a large variety of use cases; efficiency is the capability to exploit the available computational resources in order to minimize a simulation run-time. In particular, experiments show that ARGoS is capable of simulating thousands of different robots in a fraction of the time taken to an identical experiment in real time. ARGoS is currently used in 15 laboratories worldwide and in 4 EU projects, and it is released as free, open-source software.