In the past, large groups of people all heading to the same beach or city centre have been somewhat of a headache for authorities. How can you guide large crowds to their destination without a hitch? And how do you manage all those people once they reach their destination? In the Crowd Safety Manager project, the municipality of The Hague is investigating whether predictive models can help to anticipate problems. The MICD is also participating in the project with a team of TU Delft researchers.
From December 2021 to November 2022
Partners (TU Delft)
TU Delft-CEG, department of Transport & Planning
Seaside resorts like Scheveningen are usually very pleasant, but when the temperature rises and lots of people decide to take to the beach to cool off, a day at the seaside can turn into a nightmare: long traffic jams, far too few parking spaces and huge crowds sauntering down the boulevard. How can authorities help keep everyone calm and relaxed, even on busy days like these?
Road authorities, police and special investigating officers can estimate the expected traffic volume well in advance. They do so largely based on experience, while also taking into account school holidays, public holidays, scheduled events, the time of year and so on. With this information, the authorities can ensure that they have enough boots on the ground. But you also want to be able to plan ahead on busy days themselves, because when large crowds take to the streets (or the sand), a reactive approach can be sluggish and ineffective.
That is why the Municipality of The Hague has started a project with predictive data models with the aim of determining whether these models will help them proactively manage traffic flows and crowds. They also want to learn and investigate how to use these models in a proper, socially acceptable way.
In order to achieve quick results in this study, the municipality has joined forces with various private companies and the national police. Via MICD, TU Delft is also involved in the project, with TU Delft researchers contributing their knowledge of artificial intelligence, data analysis, predictive modelling and the relationship between digital technology and users.
Predictive Crowd Safety Manager
The model used in the project has been named the Crowd Safety Manager. Not only does it map the current crowdedness, but it also predicts how that crowdedness will evolve. The model draws on a wide range of variables, such as the number of visitors to the beach and the boulevard, characteristics of the area, planned and implemented crowd management measures, hotel and parking facility occupancy rates, traffic levels on access roads and so on.
Crowds, even when they are larger than expected, are not a problem per se. That is why the Crowd Safety Manager also looks at ‘aggravating circumstances’, factors such as the type of event (a fair, a dance event or a demonstration?), the prevailing sentiments and the atmosphere (determined by automatic interpretation of social media feeds), the visibility of the enforcing officers and the weather (e.g. sudden weather changes). Based on the combination of crowd levels and aggravating circumstances, the model then estimates the overall risk – and thus whether measures are necessary.
The Crowd Safety Manager is being developed and tested in the living lab in Scheveningen.