Calibration Of Pedestrian Sizes In Decision-Based Modelling

Calibration of any model is the crucial part of pedestrian movement prediction. Hence, a suitable calibration approach is needed. This paper deals with an author's microscopic decision-based model and a general calibration concept which is focused on the phase of congestion. Dense crowd behaviour is modelled by two parameters describing the agent size, namely social and physical pedestrian sizes. The physical (minimum) pedestrian size is considered to be known - it can be estimated from the real data using distance of human shoulders. However, the social size can vary during the time with respect to the surroundings of the pedestrian. Furthermore, the size estimated by a general metric may significantly differ to the individual pedestrian experience. In other words, pedestrians decide about their own (social) compression in a crowd. Therefore, the calibration process has to be based on sophisticated methods considering individual pedestrian behaviour rather than simple macroscopic quantities. Thus, this calibration study shows how pedestrian trajectories and derived microscopic quantities (especially local density) can be applied. An experimental data of passing through a room with one exit are used.