Increasing Individual Density Reduces Extra-Variance in Swarm Intelligence

Abstract

Social organisms form a swarm and forage preys, collectively and effectively [1]. The swarm has to inhibit a variance of foraging frequency for survival, which ensures stable and predictable income. In the previous study, we focused on ”trail pheromone” system which enables robots to communicate one another [2]. In the present study, we analysed statistically the variance of the foraging behaviour of the robot swarm, using repeated (48 trials) computer simulation.We set the individual density as the parameter. The density in Figures, X axis means the number of individuals on the unit field(180 × 180 [cm]). In the simulation, we set 360 × 360[cm] as the field size. When the individual density is low, the variance of the foraging behaviour is larger than expected from the random behaviour(i.e., binomial distribution; Fig. 1). Horizontal lines in Fig. 1 mean expected one from the binomial distribution. As the density goes high, the variance is reduced toward the expected value.