A Multi-Agent System to Study the Internal Displacement of Passengers and Their Distribution on a Large-Capacity Bus

Pedestrian dynamics have been widely studied in the context of crowds in panic situations. In this contribution, however, we have condensed the research undertaken on the dynamics of users inside high-capacity buses under normal conditions. The dynamics and distribution followed by users inside the bus lead to spatial distributions in which several areas present a low density and areas located close to the entrances and exits exhibit a very high density of users. This density variation is inconvenient for a number of reasons, as the capacity of the bus is not achieved because of uneven distributions, and discomfort of users is high. Through an agent-based model, we study the agents' dynamics and their interactions between them and the bus interiors such as seats and corridors. We conclude that even when users tend to move to an area in which they perceive low density of occupation, if some agents try to maintain their position close to the exit door, odd distributions are achieved. To find better policies, we tested some alternatives for users to enter and exit the bus. We found that it is possible to have a better density distribution and better comfort if other policies for the entrance/exit are implemented. Also, we applied a genetic algorithm to find seat distributions that lead to better comfort measures, subject to a given policy.