Data for Model Building and Validation for Non-lethal Weapons and Crowd Management: Initial Efforts.

摘要

There have been many efforts to create a simulation of crowd behavior using existing platforms. However, all efforts to date have been fatally flawed by the lack of crowd data for building the model, and the lack of methods and means for verification and validation of the crowd simulation. In the past two years the Target Behavioral Response Laboratory has collected multilevel crowd data and information on 200 individuals' behaviors in 15 crowd experimental runs. The basic paradigm is a rock throwing crowd facing a control force wielding a variety of simulated non-lethal weapons, including simulated hand-to-hand combat and stand-off weapons. This presentation demonstrates the theory of how to leverage this large archive of data, first to build the model, then to verify and validate crowd simulation.