Automatic design of stigmergy-based behaviours for robot swarms

Stigmergy is a form of indirect communication and coordination in which individuals influence peers by modifying the environment (e.g., pheromone trails in ants). Inspired by this mechanism, algorithms and multi-robot systems have been devised, but stigmergy-based behaviours in swarm robotics have so far been designed manually—an approach that is time-consuming, costly, hardly repeatable, and dependent on designer expertise. Here, we show that stigmergy-based behaviours can be produced via automatic design: an optimisation process based on simulations generates collective behaviours for robots that can lay and sense artificial pheromones. Experiments indicate that automatically designed behaviours are as good as—or better than—manually produced ones, and that the resulting swarms exhibit spatial organisation, memory, and communication through pheromone-based stigmergy.

Muhammad Salman, David Garzón Ramos, Mauro Birattari