dc.description.abstract | Computer models for pedestrian movement can be used to evaluate and support the evacuation designs for buildings. Therefore, the development of pedestrian models has become an important field in civil engineering. In the literature, the related studies have focused on the responses of an individual pedestrian to other pedestrians and obstacles. The more complicated interactions between pedestrians such as waiting, yielding, and detour are rarely considered. To improve the accuracy of pedestrian models, the objective of this research is to include the above behaviors in the pedestrian models in order to reproduce more realistic pedestrian movements.
In this study, the temporal and spatial relationships between pedestrians are described with time-space networks. The game theory is chosen as the decision-making mechanism for pedestrians. When an agreement exists between pedestrians, cooperative games are used to simulation pedestrian behaviors. On the other hand, competitive games are adopted if such agreement does not exist. In this study, the cooperative games are solved using mathematical programming. Because the pedestrians are inseparable, integer programming must be used to ensure that pedestrians are considered as a whole. For competitive games, K shortest path algorithm is adopted to find out each pedestrian’s strategies. An algorithm is developed to identify the strategy pairs with conflicts. Next, Pareto solutions and Nash equilibrium solutions for the pedestrian movement strategies are found. Finally, various scenarios are tested to understand the capability of the proposed approach for simulating pedestrians. Overall, the proposed simulation approach simulates pedestrian interactions under different assumptions effectively and could be useful for improving the evacuation designs for buildings.
| en_US |