A state-based multi-agent system model of taxi fleets

Abstract

Management and control of transportation systems benefit from simulation modeling. The design of the corresponding models is difficult because of their complexity. Multi-agent systems cope with this problem by a divide-and-conquer approach. However, agent model design is still quite a challenge. In this paper, we propose a layered architecture for agents where each component is a kind of a stack-based state machine of our own. This model complements extended finite-state machines with some basic state stack operations that enable not only dealing with hierarchy but also with planning, which is a key element for belief-desire-intention (BDI) agents. Special care was taken to make the representation of these extended finite-state stack machines (EFS2M) simple so that their programming is straightforward. Through an educational example we show how such class of models are, and the potentiality of the solution. The taxi fleet simulation model is a metaphor for transportation systems in structured environments like factories or warehouses but can also be used as a vehicle traffic simulator. As for the latter case, we illustrate how it can be used to determine the efficiency and the quality of service of a taxi fleet in an urban area.