Integrated Pedestrian Behavior Modeling under Normal and Evacuation Situations

Project Information

IPBMNES

Grant agreement ID: 656349

DOI 10.3030/656349

Closed project

EC signature date 22 April 2015

Start date 1 January 2017

End date 31 December 2018

Funded under

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Total cost € 171 460,80

EU contribution € 171 460,80

Coordinated by

TECHNISCHE UNIVERSITAET DRESDEN
Germany

Periodic Reporting for period 1 - IPBMNES (Integrated Pedestrian Behavior Modeling under Normal and Evacuation Situations)

Reporting period: 2017-01-01 to 2018-12-31

Summary of the context and overall objectives of the project

In the context of pedestrian behavior, the pedestrian movements are more complex and chaotic compared to vehicular flows mainly because they are not limited to the “lanes”. Moreover, in pedestrian dynamics, collective dynamics will be formed from individual human non-linear interactions. For example, a crowd in normal situations forms self-organized patterns such as lane formation, oscillatory flows at bottlenecks, and stripe formation in intersecting flows. However, under extreme conditions, the coordination will fail leading to new phenomena such as “freezing-by-heating”, “faster-is-slower”, and crowd turbulence. Table 1 shows a list of the major crowd related disasters in Europe and all over the world. The latest crowd disorder in Europe occurred on July 24, 2010 at the “Loveparade”, an electronic dance music festival in Duisburg, Germany. There 21 people died and more than 500 were injured in a stampede. Hence, there is a need for greater understanding of characteristics and behavior pattern for pedestrian in normal and evacuation process.

The following are the objectives of the proposal:

1) To develop mathematical models to understand the pedestrian crowd dynamics in mass gathering events in order to replicate the behaviour and pattern of crowd in real world conditions

2) To calibrate the model parameters using detailed pedestrian trajectory data

3) To develop pedestrian simulation software to test and design different control and management strategies and propose efficient crowd management measures during normal and evacuation situations

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

"The work has been organized into working packages:

Work Package number 1: Detailed Literature Review.\
The literature has been organized into three sections: Pedestrian behaviour, pedestrian trajectory data, and modelling approaches used. For the actual references, see the pdf of the final technical document B

Pedestrian Behaviour: The pedestrian behaviour has been subdivided into normal situations and evacuation scenarios
Pedestrian trajectory data: This includes both experiments and how to obtain trajectory data by processing the raw video material
Modelling approaches: This includes cellular automata, agent-based models, and physical social-force models.

Work Package 2: Data Collection and Extraction:

In this study, we collected large amount of pedestrian data using unmanned aerial vehicle (UAV) in different European countries such as Czech Republic, Croatia, and Serbia at different events such as music festival, football matches and historical event. The videos were recorded in 4K (4096x2160) resolution at 25 fps using the consumer quadcopter DJI Phantom 4 Pro Plus. The drone was made to fly between 40 – 60 m attitude directly above the crowd (Croatia and Serbia) and slightly away from the crowd for safety issues (Czech republic) and this would eliminate occlusion effect and minimize perspective distortions.

Work Package 3: Data transformation and smoothing techniques:
The actual trajectory data extraction by AI methods has been outsourced to the UAV operating company. However, since the provided trajectories relate to the pixel positions of the heads of the pedestrians and we need the physical coordinates of the feet, i.e. the position of the actual pedestrians, we developed, calibrated and validated an affiine-linear transformation do perform the mapping from head pixel coordinates to physical feet coordinates.
Finally, we have tested and optimized different low-pass filters to eliminate high-frequency data noise while retaining the actual information

Work Package 4: Model Development and Calibration:
Models are essential in order to describe the crowd movement in a realistic way. In microscopic models, the actions, decisions and behaviours of each pedestrian as well as interactions with others are considered. We select, adapt, calibrate, and validate a variant of the social-force model (""elliptical specification"")

Work Package 5: Pedestrian Simulation and Crowd Management Strategies:
Most of the human stampedes are caused by a combination of many factors such as high crowd density, local reduction of the capacity (bottleneck), panic, etc. Therefore, an in-depth understanding of the interactions between pedestrians during normal and evacuation situations is needed for devising better crowd control and management strategies. Microscopic simulation is an appropriate tool for investigating this since it allows us to treat ‘pedestrians’ individually and their interactions. We are now in the process of developing the software. The simulation model comprises of three major components: the pedestrians, the environment (in which pedestrians are situated) and the behaviour rules (to identify how pedestrians and other objects interact with each other). Each pedestrian registers the position of obstacles, current position of other pedestrians including his/her perception about their speed and the direction of movement."

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

The simulation application to be developed as a part of this study can be a powerful tool for investigating and evaluating different management strategies with respect to pedestrian safety. The results and findings from this study will find applications in developing early warning system and crowd management at mass gatherings. This project is beneficial for local authorities, planners and event managers to assess suitable dynamic crowd control measures during normal and evacuation situations.

Pedestrian crowd at a historical event in Seljacka Buna, Croatia

Pedestrian crowd at a football stadium in Ostick, Croatia

Pixel markers indicating the heads as delivered by the UV company

Another UAV snapshot of the historical event Seljacka Buna, Croatia

Trajectories of pedestrians at the Seljacka Buna, Croatia