Anticipatory Vehicle Routing in Stochastic Networks Using Multi-Agent System

In this paper, we consider an anticipatory vehicle routing problem. We investigate the architecture of a multiagent routing system wherein each vehicle is represented by an agent that interacts with the road infrastructure. We use a route reservation approach, whereby each vehicle reserves a space-time slot on the road segments included in the route. It makes it possible to predict the load of the segments and estimate the travel time with higher accuracy. To estimate the travel time, it is proposed to take into account the stochastic properties of the network. In this paper, we consider not only the average travel time but also its variance. The proposed approach is compared with alternative routing strategies: uncontrolled routing and anticipatory routing in a deterministic transport network. We evaluate the proposed model using the microscopic simulation of a real-world traffic environment with SUMO modeling software.