Experiments And Simulations Of Pedestrian Walking And Evacuation Behaviors Under Several Scenarios

Walking is the most elementary and primitive travel mode,also is an integral part of people's various travel activities.Unlike vehicle's movements,which are subjected to fixed corridors of road environment as well as specific traffic rules,pedestrian movements are distributed over the whole urban areas,and are more flexible and stochastic without restrictions of specific rules.Moreover,pedestrian's walking behaviors are repeating and evolving day to day in various scenarios and conditions.Therefore,they are complicated.It is not able to predict pedestrian behaviors in various scenarios based on simple stimulus-response patterns.For researchers,to better recognize and understand complicated pedestrian behaviors,it is necessary to perform extensive field observations and controlled laboratory experiments to investigate and master the complicated,nonlinear and dynamical interactions among pedestrians and between pedestrians and their surrounding environment.Then,building a simulation model to reproduce the observed behaviors and extending it to a broader set of scenarios to predict pedestrian behaviors.Thus,it is possible to provide theoretical basis and scientific instructions for designing and assessing pedestrian facilities,as well as managing the large-scale crowd evacuation.Following this,in this thesis,pedestrian behaviors are studied under various scenarios based on controlled laboratory experiments.Furthermore,related microscopic simulation models are proposed to model and analyze pedestrian behaviors.Specifically,the contents of this dissertation mainly include the following four aspects:(1)Experimental study and analysis of pedestrian counter flow in discrete space and time.Cellular automaton(CA)has been widely used in modelling pedestrian movement.However,one problem of CA approach was that deadlock usually occurred in pedestrian counter flow.One possible reason was that existing CA models still do not correctly represent the cognitive and decision-making processes of real people.To investigate whether the deadlock would occur when real people were required to walk in discrete space and time,an experimental study on pedestrian counter flow was performed,in which participants were required to walk in discrete space and time.Remarkable cooperation behaviors of pedestrians have been observed,which enable the formation of exit rows in the counter flow and prevent the formation of deadlock.Furthermore,a comparative experiment was performed in which pedestrians walked under normal condition.The comparative studies revealed that while results of the two experiments were similar at low densities,significant difference occurred at high densities.The experimental results indicated that to fully describe pedestrian counter flow with CA approach,the remarkable cooperation behaviors of pedestrians need to be considered.Moreover,from the comparative results,one might also need to consider other factors such as flexible characteristics of pedestrians,small cell size,and various walk speeds.(2)A behavior based cellular automaton model of pedestrian counter flow.A behavior based CA model for pedestrian counter flow was presented.The behaviors of active slowing down and lane changing are considered to simulate the evolution of pedestrian counter flow in a corridor with open boundaries.A concept of dominant row was introduced to depict pedestrian's lane changing behavior.The velocity profile has been investigated and two separation times have been studied,which quantify the evolution process of lane formation in pedestrian counter flow.Simulation results indicated that steady separate lanes can form even in a dense condition.Therefore,the gridlock formation has been remarkably suppressed compared with some typical CA models for pedestrian counter flow.(3)The impacts of communication on pedestrian evacuation in view and hearing limited condition.Pedestrian evacuation in view and hearing limited condition was investigated based on the social force approach.It was assumed that there were two types of pedestrians composing the crowd:informed individuals know the exit location whereas uninformed individuals do not.The uninformed individuals could communicate with the informed ones within their perceptual fields during the evacuation,thus learnt to know and memorize the exit location.The cases with and without communication/memory were considered.The simulations indicated that communication and memory were able to enhance the evacuation efficiency.The influence of communication between the informed and the uninformed individuals on the utilization of an emergency exit was also investigated.(4)Experimental study and modelling of pedestrian wall-following behavior under limited visibility condition.Wall-following behavior is believed to be of great importance to orient and navigate pedestrian's way-finding during an emergency evacuation,under the circumstance that the visibility and hearing are limited.Though it has been considered in many pedestrian evacuation models,the empirical and experimental results on the wall-following behavior were scarce.The questions in terms of how pedestrian approach the wall,how they decide the wall-following direction(clockwise or anticlockwise),and how they deal with conflicts in the wall-following process were still not well understood.To this end,an evacuation experiment was performed in a mock room.Thirty college students were recruited to participate in the experiment.Each participant wore a baseball cap covered with an opaque veil to create a limited visibility environmental condition.Thirteen runs of the experiment have been performed.The behavioral pattern of each participant with a total of 270 cases was recorded and studied.Based on these findings,the above three questions were answered.Furthermore,a wall-following model in the framework of the social force model,considering the experimental findings.The simulation results of pedestrian evacuation process are consistent with the experimental outcomes.