Coupling Counter Flow Model And Side-shoulder Model Of Social Force Model And The Experimental Calibration Of Its Parameters

In recent years,with the development of economy and the improvement of urbanization degree,more and more people are settling down in the city,which makes the urban public space need to accommodate more and more people.When many people move in the public place,individual movement will be affected by the movement of the surrounding crowd,and there is a subjective tendency to avoid collision,so their movement behavior is complicated.In the process of pedestrian movement,there are both translational movement and rotation,namely the "turning" behavior of pedestrians:in wide space,when facing the oncoming crowd,pedestrians will constantly choose and adjust their direction of movement to avoid a frontal collision;in narrow spaces,pedestrians will rotate their shoulders to avoid collisions with other pedestrians moving in opposite directions.In order to simulate the collision avoidance behavior of people when they meet the reverse crowd in a wide or narrow area,scholars developed the counter flow model and the side-shoulder model respectively,which are essentially the direction selection during translational motion or the side-shoulder behavior during rotation.In the actual crowd movement,these two kinds of collision avoidance behaviors often work together,so it is necessary to simulate or predict the behavior of individuals rotating while choosing the direction.In this paper,an improved social force model considering direction choice and turning behavior is proposed by introducing mixed coefficient on the basis of proper modification of the original counter flow model and side-shoulder model.In addition,in order to determine the value of mixing coefficient,this paper proposes a spatial coordinate conversion formula based on monocular vision.The pixel coordinates of two joints on the left and right shoulder and larynx of pedestrians can be obtained by Open Pose,and the world coordinates of the spatial position and rotation angle of pedestrians in the process of movement can be obtained by using the coordinate conversion formula.After comparing and analyzing the actual pedestrian data,the reasonable value of mixing coefficient is determined.The research content of this paper is as follows:First of all,in order to obtain the image coordinates of pedestrians,this paper uses the Open Pose algorithm to detect the joint nodes and skeleton of pedestrians,so as to obtain the pixel coordinates of the two joint nodes of the left and right shoulder of pedestrians and the throat required by the coordinate transformation,based on which the positioning features can be constructed.In order to obtain the world coordinates of pedestrians,this paper uses the proposed coordinate transformation formula to convert the pixel coordinates of pedestrians into world coordinates so as to achieve real spatial positioning.Through the experiment: measured the position information of eight relatively small objects relative to the projection point of the camera on the ground,and obtained the photos of the objects through the fixed-point shooting method to obtain the pixel coordinates of the objects,and then obtained the calculated value of the object coordinates through the coordinate conversion formula.By comparing the real and calculated values of the coordinates,we can see that the accuracy of the proposed coordinate transformation method is more than 95%.Therefore,this localization method not only has high accuracy,but also is simple and low cost,which provides an experimental basis for the subsequent research.Then,this paper proposes a coupling counter flow model and side-shoulder model of social force model,which combines the counter flow model with the side-shoulder model on the basis of the social force model,so as to improve the social force model.The model can be applied to open Spaces rather than narrow passages.In the mixing coefficient of 0.1,0.3,0.5,0.7 and 0.9,0.3 is determined as the final mixing coefficient.At the same time,the correctness of the mixed model is verified by comparing with the experiment.Finally,by comparing the simulation results of special scenes,it is found that the coupling counter flow model and side-shoulder model of social force model can better reproduce the direction choice behavior and side-shoulder behavior of pedestrians in the actual process of movement,so as to avoid the phenomenon of pedestrian collision contrary to the actual situation in the simulation.In conclusion,the aim of this paper is to propose a more realistic crowd countercurrent model that takes into account individual direction selection and rotational behavior during countercurrent.The pedestrian location method proposed in this paper not only has high accuracy,but also is simple and low cost,which provides an experimental basis for the subsequent research.By introducing mixing coefficient to improve the social force model,the mixed model proposed in this paper can more accurately describe the individual movement under the countercurrent condition.The simulation results reveal the authenticity of the coupling counter flow model and side-shoulder model of social force model can more accurately describe pedestrian movement and provide a theoretical basis for the subsequent research on crowd movement and evacuation.