Robust Crowd Control and Performance Evaluation of Scenario Generation

Abstract

We propose a robustness evaluation method for crowd control produced by expectation optimization to relieve the congestion at large-scale events. Crowd control, such as route guidance and destination allocation, can relieve the congestion. The people on site often determine crowd control in practice but the combination of crowd simulation and stochastic optimization can seek robust crowd control that is adaptable to many situations. However, the effectiveness of crowd control strategies derived from these techniques has not been sufficiently verified in unknown situations. This study proposes the robustness evaluation method for crowd control in unknown situations. The proposed method evaluates the robustness of crowd control based on the difference between the evaluation values of solutions obtained under complete information and incomplete information. In the experiments, We evaluate the robustness of gate allocation when exiting a large-scale event venue and verify the performance of an expectation optimization and scenario generation method that defines an uncertainty set in expectation optimization.