Exploiting bacterial swarms for optimal coverage of dynamic pollutant profiles

Inspired by the simplicity of how nature solves its problems, an approach that would enable robotic agents form a visual representation of an invisible hazardous substance is presented. Such an approach would be very useful if emergency services need to systematically and strategically evacuate an area affected by invisible substance especially in a situation where evacuation resources are limited. The approach presented is computationally cheap and yet highly effective when compared to other methods such as Voronoi partitioning or simulated annealing. In addition, it does not require a polygon derived environment or a prior knowledge of the environment or pollutant and can be used to trace dynamically changing pollutant profiles. The experimental results prove that the algorithm presented in this work always converges to a pollutants distribution even though the pollutants profile is of a complex nature with no clear gradient boundaries.