The understanding and prevention of catastrophes at large-scale events are of utmost societal importance. For that, pedestrian real-time simulations would be a potent tool. In this thesis, I introduce parallelism to optimal steps models and develop efficient and parallel algorithms to construct so-called navigation fields. A new meshing algorithm reduces the problem size and a novel numerical method exploits similarities of consecutively solved eikonal equations. In combination, real-time pedestrian simulation becomes possible for many large-scale scenarios.
«
The understanding and prevention of catastrophes at large-scale events are of utmost societal importance. For that, pedestrian real-time simulations would be a potent tool. In this thesis, I introduce parallelism to optimal steps models and develop efficient and parallel algorithms to construct so-called navigation fields. A new meshing algorithm reduces the problem size and a novel numerical method exploits similarities of consecutively solved eikonal equations. In combination, real-time pedest...
»