In this paper we develop an approach to modeling and simulating the process of infection transmission among individuals and the effectiveness of protective counter-measures. We base our approach on pedestrian dynamics and we implement it as an extension of the Vadere simulation framework. In order to enable a convenient simulation process for a variety of scenarios, we allow the user to interact with the simulated virtual environment (VE) during run time, for example, by dynamically opening/closing doors for room ventilation and moving/stopping agents for re-positioning their locations. We calibrate and evaluate our approach on a real-life case study—simulating COVID-19 infection transmission in two kinds of scenarios: large-scale (such as the city of Münster, Germany) and small-scale (such as the most common indoor environments—classrooms, restaurants, etc.). By using the tunable parameters of our modeling approach, we can simulate and predict the effectiveness of specific anti-COVID protective measures, such as social distancing, wearing masks, self-isolation, schools closing, etc.

Crowd simulations are widely used to study and predict the human behavior in disaster scenarios. In this paper, we introduce real-time user interactivity into the simulation process of virtual environments (e.g., buildings with rooms and doors between them). We develop a new tactical path-planning model that translates the interactive virtual environment into an abstract graph in order to calculate the shortest paths in real time. Our extension of the Vadere simulation framework with interactivity features allows the users to better understand the actual problem situations and to analyze them. Our experiments demonstrate the effectiveness of the approach by simulating the evacuation of students in groups and as individuals from the Schloss Muenster (the administrative building of the University of Muenster) in Germany. During simulation run time, the user can interact with the virtual environment spontaneously (e.g., by opening and closing doors) while our model recalculates the shortest paths for agents in real time.