Improvements in the design of pedestrian facilities have reduced the frequency of crowd accidents, and safety is now generally ensured in well-planned crowd events. However, congestion and inefficient use of infrastructures still remain an issue. To guarantee comfort and reduce close contacts between people, there are circumstances when crowd density may have to be reduced well below safety limits. Although research has given a lot of attention to extreme scenarios, methods to improve non-critical conditions have been little explored. In addition, crowd sensing technology is still mostly used for data collection and direct use on crowd management is rare. In this work, we present a system aimed at computing optimal egress time for groups of people leaving a complex facility. We show that, if egress starting time is accurately computed for each group based on actual crowd conditions, density can be greatly reduced without having a large effect on the total egress time of the whole crowd. To show the efficacy of such a system, a small-scale experiment is conducted where all components are tested in a simple scenario. As a result, an increase in total egress time by only 5% allowed to reduce maximum density by 35%.
Keywords: crowd dynamics; dynamic scheduling; egress; pedestrian sensing; pedestrian traffic; real-time simulation.
© 2020 The Authors.