MO-LOST: adaptive ant trail untangling in multi-objective multi-colony robot foraging

摘要

In the context of large-population multi-objective robot foraging, we present a novel ant-inspired trail-following algorithm that is able to adaptively untangle multiple trails. The emergent result is often a set of short, non-intersecting trails that produce good system throughput due a good trade off between the dual goals of minimizing travel distance and spatial interference. Empirical simulation experiments with up to 200 robots suggest that the method can usefully improve performance in practice.