Directional ambiguity in trail-laying algorithms

We developed multi-agent models inspired by ant foraging. Previous research shows that ants obey Weber's law and change their moving angles based on pheromone conditions. We constructed trail-laying models in which agents estimate the global pheromone gradients based on local pheromone gradients, which resulting in emergent Weber's law. Agents were also admitted to use directional cues based on motions of other local agents to produce realistic curved paths. We also showed that re-estimation of pheromone gradients based on directional cues yields a balance between exploitation and exploration.