Computational Modeling

In the field of pedestrian simulation, two aspects are particularly important: How do I move in space and when do I move, i.e. spatial and temporal decisions. Here we distinguish between a discrete up to a quasi-continuous mapping:

Why quasi-continuous? Since each computer can calculate only discretely (each calculation is executed in the clock of the processor) a temporal discretization takes place inevitably for each computer program of the world. The same is true for the spatial discretization: The continuity existing in reality can only be approximated by the computers.

However, the degree of resolution varies. While some models resolve the geometry cell-based to agent size, i.e. are still quite coarse, there are other models (such as crowd:it) whose standard resolution is quasi-continuous by interpolating support points 10cm away. The setting of the grid points can be adjusted individually and increased simply by clicking.

For the mapping of the temporal decision there are also different accuracies. Either the model assigns all agents a fixed time in which each is allowed to take a step (usually several times per second) or an agent can decide to leave at any time. The former case is implemented in models based on differential equations. These equations are then re-evaluated at each time point (e.g., social force models). We have now changed this decision from synchronized time steps to individual time steps. Because this is how reality works: people decide when to take a step without a globally ticking clock. In this way, we manage to represent human behavior even better and to get more reality into the simulation. Our software can transfer another aspect from reality into the model world.