Selection of results for four maps of the simulation.

All maps are delimited with perimeter walls. The first map (Id 1) is an empty hall of 100x100m (left map of Fig 3a). Map 2 is a 100x200m empty hall. Map 3 is half divided hall with one wall and only a pass door. Map 4 is a complex 100x100m hall, with multiple walls and corridors (middle right map of Fig 3a). The entropy for each map ordered by increasing Id is (0.2405, 0.0027, 0.2817, 0.4639). The average of area covered for 10 runs and the average of time required to cover this area is shown. The maximum time allowed for simulation is 15000s. This time will be shorter if the swarm cover (using the ca metric) the 95% of the area. Furthermore, the number of individuals in states wander and beacon w, b for the simulation and for the mathematical model is presented. Two different covering metrics are shown: ca, the percentage of terrain that is covered by any individual of the swarm in any time and cb, that indicates the percentage of area covered by individuals in the final iteration of the behaviour, assuming a detection range of 10m per individual. Finally, the measure of the differences between the number of individuals in each state predicted and simulated (using root-mean-square deviation, RMSD) is also presented.