Periodic Reporting for period 1 - iMTFM (Intelligent Multi-modal Transfer Flow Management system)
Reporting period: 2020-08-01 to 2022-01-31
The iMTFM aims to provide support for decision-makers (i.c. Station Managers) to identify and predict risky situations. Risk in this case pertains to a range of conditions that can be harmful to the traveller. For stations, these risks pertain to crushing due to overcrowding (e.g. in case of special events or disturbances), risks of falling (e.g. near escalators) in case the station or the platform is overcrowded, and the risk of falling off the platform.
The pandemic has forced the team to reconsider the approach to ensure that the methods developed are applicable to various circumstances. This led to the development of the so-called Bowtie method, in which on the one hand we focus on the collection of data, and the estimation and prediction of (objective) crowdedness on the relevant locations. On the other hand, we identify 'aggravating' circumstances (e.g. based on prevailing sentiments of the crowd, weather conditions, characteristics of the population, and trains passing by). The objective predictions of crowding are combined with these aggregating factors to provide an assessment of (predicted) risk, which is provided to the decision-maker.
While work is still continuing, the proposed methodology has already been deployed in various projects both inside and outside of the railway station context. For one, together with ProRail and Arane, we developed a methodology to identify risky situations on station platforms using in order to determine which stations in The Netherlands require platform redesign interventions. Second, we have deployed the Bowtie concept in the Crowd Safety Manager Scheveningen (together with the company Argaleo). Here, we have developed novel AI methods to forecast crowdedness (up to six days ahead). This system is used by decision-makers to plan possible interventions to ensure the safety of visitors and residents in Scheveningen, The Hague.
The pandemic has forced the team to reconsider the approach to ensure that the methods developed are applicable to various circumstances. This led to the development of the so-called Bowtie method, in which on the one hand we focus on the collection of data, and the estimation and prediction of (objective) crowdedness on the relevant locations. On the other hand, we identify 'aggravating' circumstances (e.g. based on prevailing sentiments of the crowd, weather conditions, characteristics of the population, and trains passing by). The objective predictions of crowding are combined with these aggregating factors to provide an assessment of (predicted) risk, which is provided to the decision-maker.
While work is still continuing, the proposed methodology has already been deployed in various projects both inside and outside of the railway station context. For one, together with ProRail and Arane, we developed a methodology to identify risky situations on station platforms using in order to determine which stations in The Netherlands require platform redesign interventions. Second, we have deployed the Bowtie concept in the Crowd Safety Manager Scheveningen (together with the company Argaleo). Here, we have developed novel AI methods to forecast crowdedness (up to six days ahead). This system is used by decision-makers to plan possible interventions to ensure the safety of visitors and residents in Scheveningen, The Hague.