We present a novel multi-robot simulator named ARGoS. ARGoS is designed to simulate complex experiments involving large swarms of robots of different types. ARGoS is the first multi-robot simulator that is at the same time both efficient (fast performance with many robots) and flexible (highly customizable for specific experiments). Novel design choices in ARGoS have enabled this breakthrough. First, in ARGoS, it is possible to partition the simulated space into multiple sub-spaces, managed by different physics enginesrunning in parallel. Second, ARGoS' architecture is multi-threaded, thus designed to opti- mize the usage of modern multi-core CPUs. Finally, the architecture of ARGoS is highly modular, enabling easy addition of custom features and appropriate allocation of compu- tational resources. We assess the efficiency of ARGoS and showcase its flexibility with targeted experiments. Experimental results demonstrate that simulation run-time increases linearly with the number of robots. A 2D-dynamics simulation of 10,000 e-puck robots can be performed in 60 % of the time taken by the corresponding real-world experiment. We show how ARGoS can be extended to suit the needs of an experiment in which custom functionality is necessary to achieve sufficient simulation accuracy. ARGoS is open source software licensed under GPL3 and is downloadable free of charge
ARGoS: a modular, parallel, multi-engine simulator for multi-robot systems
Trianni Vito;
2012
Abstract
We present a novel multi-robot simulator named ARGoS. ARGoS is designed to simulate complex experiments involving large swarms of robots of different types. ARGoS is the first multi-robot simulator that is at the same time both efficient (fast performance with many robots) and flexible (highly customizable for specific experiments). Novel design choices in ARGoS have enabled this breakthrough. First, in ARGoS, it is possible to partition the simulated space into multiple sub-spaces, managed by different physics enginesrunning in parallel. Second, ARGoS' architecture is multi-threaded, thus designed to opti- mize the usage of modern multi-core CPUs. Finally, the architecture of ARGoS is highly modular, enabling easy addition of custom features and appropriate allocation of compu- tational resources. We assess the efficiency of ARGoS and showcase its flexibility with targeted experiments. Experimental results demonstrate that simulation run-time increases linearly with the number of robots. A 2D-dynamics simulation of 10,000 e-puck robots can be performed in 60 % of the time taken by the corresponding real-world experiment. We show how ARGoS can be extended to suit the needs of an experiment in which custom functionality is necessary to achieve sufficient simulation accuracy. ARGoS is open source software licensed under GPL3 and is downloadable free of chargeI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
