Scalable Signal-Based Simulation of Autonomous Beings in Complex Environments

Simulation of groups of autonomous beings poses a great computational challenge in terms of required time and resources. The need to simulate large environments, numerous populations of beings, and to increase the detail of models causes the need for parallelization of computations. The signal-based simulation algorithm, presented in our previous research, prove the possibility of linear scalability of such computations up to thousands of computing cores. In this paper further extensions of the signal-based models are investigated and new method for defining complex environments is presented. It allows efficient and scalable simulation of structures which cannot be defined using two dimensions, like multi-story buildings, anthills or bee hives. The solution is applied for defining a building evacuation model, which is validated using empirical data from a real-life evacuation drill.