Spatial knowledge modelling for agent based system

Efficient simulation of population movement in urban networks is crucial for urban planning, traffic management, and emergency response systems. In this context, we are interested in the shortest path problem in a multi-agent system. In this paper, we present a multi-level model that distinguishes between two levels of granularity: the network reality level and the agent level. For this, we use graphs representing urban networks, with primal modeling, whose edge weights can vary according to the agents and evolve over time. To achieve this, we employed the Customizable Contraction Hierarchies method, which applies multi-phase preprocessing to speed up shortest-path queries and weight updates. Experimental results show that the CCH method significantly improves computational efficiency in shortest path queries.