Football as an exact science: how ITMO University helps the organizers of the Confederations Cup and the 2018 World Cup
The Confederations Cup comes to an end smoothly - only a couple of decisive matches remain. However, football is not only a beautiful game, but also an occasion for serious scientific research. In this material we will describe how ITMO University scientists work on projects that help make watching a football game safer and more comfortable.
/ Photo by Rama V / CC
This is a project of the Institute of High Technology Computer Technologies (NII NKT) of the ITMO University, which allows to model and detail the behavior of large crowds of people in different conditions. One of the issues that are being worked on at NII NKT is predictive modeling of complex systems. In this case, complex are understood as systems with a large number of elements (connected by the principle “every with each”), which are in the precritical state - when small changes in properties can lead to a change in the system as a whole.
')
The research and development project of the NKT on modeling crowd behavior allows forecasting the development of events in crowded places — and taking into account features of the place, traffic intersections, weather conditions, as well as losing various scenarios, from sudden downpour to hitting your favorite team or shouting about the explosion.
All of these scenarios can become reality - we are talking about situations that threaten the lives and health of people. A recent example is the stampede when watching the Champions League final in Turin on Piazza San Carlo on the third of June of this year. Predictive modeling helps to understand how to avoid such moments, how to properly respond to event organizers, how to correctly design objects of a mass gathering of people and the surrounding infrastructure.
One of the authors of the project, an employee of the Institute of High Technologies Vladislav Karbovsky, explains the tasks and functions of the project to the journalist Izvestia :
In their work, scientists use multi-agent modeling techniques. Artificial intelligence is responsible for the behavior of each “participant” in the model, and the actions of the entire flow of people are determined taking into account the trajectory and speed of movement of individual objects. The founder and director of the Institute, Professor Alexander Bukhanovsky told the Russian Newspaper about this :
As part of a computer simulation, scientists can combine several models at once (the environment for combining multi-agent models was created by the staff of ITMO University together with colleagues from universities in Mexico, the Netherlands and the USA).
For example, you can combine a model that describes the behavior of people in an emergency situation on a given object, and a traffic model within this object. For their "gluing together" I had to write a special software - the already existing analogues of scientists were not satisfied. As a result, it became possible to simulate various scenarios of crowd behavior depending on the changing context and research tasks.
The calculations are complicated by the chaotic movement of the crowd: the model takes into account such factors as changing the object's point of interest, choosing the desired speed, unexpected stops, and so on. To obtain reliable results, virtual experiments must be repeated many times with different inputs.
This analysis was carried out by scientists from ITMO University, in particular, in relation to the St. Petersburg stadium "Zenit-Arena" - they studied the scenarios of the movement of football fans after the match:
The project team paid particular attention to logistics issues: prior to the start of the Confederations Cup, a study was conducted on the movement routes of fans both within St. Petersburg and throughout the country. At the same time, various scenarios were calculated, including the “transport collapse” scenario, which, fortunately, was avoided.
Before building a model, scientists use real-life information about actions and movements of people. The initial data will be the recordings of video surveillance cameras, information of turnstile systems, information on the density of distribution of mobile subscribers.
In addition to experimental data on the movement of people, scientists use information from social networks. This allows you to assess the mood of the crowd and predict the likelihood of fights and the emergence of aggressive fans:
/ Vasyl Boichuk talks about the developments of the ITMO University for the TV program "Project 2015"
Such modeling is important for decision making in both the long and the short term. It helps to plan in advance the details of mass events (for example, it is important for the organizers of a football match to know how to form the optimal configuration of the barriers). And also allows you to make the right decision in the event of abrupt changes: for example, "lose" the situation with sudden showers or temperature fluctuations.
In addition, the simulation allows you to work out countermeasures in case of emergency situations (including a massive panic) and to identify in advance "bottlenecks" in the system. Scientists emphasize that their task is to make recommendations for decision-making based on accurate forecasts and identifying patterns in the behavior of complex systems. And also (if possible) to simulate the "butterfly effect" - a situation in which a change in one of the minor factors leads to fundamental changes in the behavior of the system.
This work not only helps to plan events, but also contributes to the effective development of urban spaces. ITMO University’s developments will also be used during security at the 2018 FIFA World Cup.
/ Photo by Rama V / CCPredictive modeling of complex systems
This is a project of the Institute of High Technology Computer Technologies (NII NKT) of the ITMO University, which allows to model and detail the behavior of large crowds of people in different conditions. One of the issues that are being worked on at NII NKT is predictive modeling of complex systems. In this case, complex are understood as systems with a large number of elements (connected by the principle “every with each”), which are in the precritical state - when small changes in properties can lead to a change in the system as a whole.
')
The research and development project of the NKT on modeling crowd behavior allows forecasting the development of events in crowded places — and taking into account features of the place, traffic intersections, weather conditions, as well as losing various scenarios, from sudden downpour to hitting your favorite team or shouting about the explosion.
All of these scenarios can become reality - we are talking about situations that threaten the lives and health of people. A recent example is the stampede when watching the Champions League final in Turin on Piazza San Carlo on the third of June of this year. Predictive modeling helps to understand how to avoid such moments, how to properly respond to event organizers, how to correctly design objects of a mass gathering of people and the surrounding infrastructure.
One of the authors of the project, an employee of the Institute of High Technologies Vladislav Karbovsky, explains the tasks and functions of the project to the journalist Izvestia :
The main objective of the program is to increase security during global mass events. It is possible to predict, for example, what will happen in the event of a panic, where potentially dangerous areas are in the event of a crush, what happens if one of the entrances has many visitors. [...]
The program introduces data on the simulated building or infrastructure, the rules for moving people. At the exit - the forecast of the dynamics of pedestrian flows. You can predict scenarios: "what will happen if ..."
- Vladislav Karbovsky
In their work, scientists use multi-agent modeling techniques. Artificial intelligence is responsible for the behavior of each “participant” in the model, and the actions of the entire flow of people are determined taking into account the trajectory and speed of movement of individual objects. The founder and director of the Institute, Professor Alexander Bukhanovsky told the Russian Newspaper about this :
To build a computer model of the behavior of fans, you first need to make equations that describe the behavior of the individual. [...] And behavior in extreme situations. For example, during rain, people try to move away from each other, and the space around them increases.
- Alexander Bukhanovsky
As part of a computer simulation, scientists can combine several models at once (the environment for combining multi-agent models was created by the staff of ITMO University together with colleagues from universities in Mexico, the Netherlands and the USA).
For example, you can combine a model that describes the behavior of people in an emergency situation on a given object, and a traffic model within this object. For their "gluing together" I had to write a special software - the already existing analogues of scientists were not satisfied. As a result, it became possible to simulate various scenarios of crowd behavior depending on the changing context and research tasks.
The calculations are complicated by the chaotic movement of the crowd: the model takes into account such factors as changing the object's point of interest, choosing the desired speed, unexpected stops, and so on. To obtain reliable results, virtual experiments must be repeated many times with different inputs.
This analysis was carried out by scientists from ITMO University, in particular, in relation to the St. Petersburg stadium "Zenit-Arena" - they studied the scenarios of the movement of football fans after the match:
The project team paid particular attention to logistics issues: prior to the start of the Confederations Cup, a study was conducted on the movement routes of fans both within St. Petersburg and throughout the country. At the same time, various scenarios were calculated, including the “transport collapse” scenario, which, fortunately, was avoided.
How to collect information for analysis
Before building a model, scientists use real-life information about actions and movements of people. The initial data will be the recordings of video surveillance cameras, information of turnstile systems, information on the density of distribution of mobile subscribers.
“The most difficult thing when modeling panic is to obtain experienced data on patterns of behavior and the probability of choosing escape routes”
- Daniel Voloshin, Engineer of Scientific Research Institute
In addition to experimental data on the movement of people, scientists use information from social networks. This allows you to assess the mood of the crowd and predict the likelihood of fights and the emergence of aggressive fans:
We collect information before the match, during and after the match. Most of the data is collected before and during the match. We use geo-referencing. We collect data that people publish on Vkontakte and Instagram, and based on this information, a forecast is built up: will the flare be lit in the stands during a match, will there be a fight. To intervene quickly and prevent these extreme situations.
- Engineer NII NKT Vasily Boichuk
/ Vasyl Boichuk talks about the developments of the ITMO University for the TV program "Project 2015"
Why is it important
Such modeling is important for decision making in both the long and the short term. It helps to plan in advance the details of mass events (for example, it is important for the organizers of a football match to know how to form the optimal configuration of the barriers). And also allows you to make the right decision in the event of abrupt changes: for example, "lose" the situation with sudden showers or temperature fluctuations.
In addition, the simulation allows you to work out countermeasures in case of emergency situations (including a massive panic) and to identify in advance "bottlenecks" in the system. Scientists emphasize that their task is to make recommendations for decision-making based on accurate forecasts and identifying patterns in the behavior of complex systems. And also (if possible) to simulate the "butterfly effect" - a situation in which a change in one of the minor factors leads to fundamental changes in the behavior of the system.
This work not only helps to plan events, but also contributes to the effective development of urban spaces. ITMO University’s developments will also be used during security at the 2018 FIFA World Cup.
Source: https://habr.com/ru/post/331986/
All Articles