(234e) A Study of the Dynamics of Human Pedestrians Using Experiments and Simulations in the Indian Context

The world population, currently at about 7.5 billion, has experienced dramatic growth rates since the industrial revolution. We are poised to add another 2 billions by 2050, with a significant portion residing in the developing countries. Even with the existing numbers, India and other developing countries are witnessing unprecedented levels of population density. The management of these large volumes of crowds, especially in mass gatherings and public spaces, is an extremely challenging problem. One should also remember the fact that, in many cases, several well-populated public spaces are interconnected with a steady flow of crowds between them, which creates a network and further complicates the problem. Quite unsurprisingly, we witness a significant number of disasters like stampede, accompanied by losses of hundreds of lives every year. To prevent such unfortunate scenarios, we need to optimally design public spaces such that the chances of these disasters are minimized. These should ideally incorporate the fundamental nature of interactions, in a mathematically usable form, between human beings in such scenarios. Through significant improvements over the past decade, researchers have made significant progress towards obtaining physical models for the forces between human beings. There are several such ‘social force’ based models for human pedestrian interactions that have been proposed in literature. However, none of these were parameterized with existing scenarios of a developing country like India, where the population densities are an order of magnitude larger relative to the western nations. Higher densities, coupled with cultural and social differences, could give rise to significantly different forces between pedestrians in India. To the best of our knowledge, no such studies exist in the Indian context.

In this study, we try to understand and model the pedestrian interactions in Indian public spaces. We perform experiments in the “ultra-dilute” regime, which consist of a pair of pedestrians moving towards each other, to understand the validity of the existing ‘social force’ models in the Indian context. Also, we perform some experiments on single isolated pedestrians, which provided us with the characteristics of the motion in the absence of other interactions. In these experiments, the pedestrians mostly consisted of student volunteers at IIT Kanpur. We employ standard image processing techniques to obtain the trajectories of the pedestrians for further analysis. These trajectories provide the position, velocity and acceleration of the pedestrians at all times during the experiment. Fourier transforms of the data from experiments reveal the dynamics at various length and time scales, for both the single and a pair of pedestrians. For comparison of these with the predictions of the ‘social force’ models, we perform simulations of a pedestrian pair moving towards each other. These simulations modelled each pedestrian as a particle, which exert forces upon each other that are estimated from the relevant social force model. In the process, we also compare the predictions of two promising social force models existing in literature to highlight the similarities and differences. Further, by comparing the dynamics obtained from simulations to our experiments with volunteers, we identify some severe limitations of these social force models. Some of the qualitative trends obtained from the experiments, especially the velocity and accelerations of the pedestrians as they approach each other, are in sharp contrast with those from simulations. Our observations from experiments are quite different from the fundamental characteristics that the social force models are expected to capture, thereby raising doubts about whether they can be used in the current form for Indian scenarios.