Monday, 8:30 a.m., somewhere in France…Crowds of people are waiting on the metro platform. A packed train pulls into the station. It’s rush hour and passengers jostle to get on the train. They don’t intend to wait for the next one.

It’s a scene played out most days in big city metro stations – and one of the five main sources of frustration with public transport identified by the Keoscopie mobility trends observatory. It has even become a source of stress since the COVID pandemic, due to social distancing protocols. Today, crowding is a key factor for passengers when choosing how to reach their destination, which is why both operators and PTAs are working to anticipate it better.

Several cities have grasped the nettle of crowding to avert nightmare scenarios. How can we anticipate peak usage? How do we help passengers choose the best options before setting out? Increasingly, transport professionals are innovating with passenger information and travel demand forecasting solutions. In Dijon, for example, Keolis Dijon Multimodalité has installed signage to guide students towards bus and tram stops in the city’s main square. The goal is to ease crowding in trams at peak hours by encouraging students alighting from line T2 to catch the Liane 3 bus to the university campus rather than line T1, with almost the same trip time. This approach has proved popular, earning Keolis Dijon an award from the French transport innovation agency.

In Besançon, to anticipate peak overcrowding Keolis Besançon Mobilités has been trialling its Ginko app since last September with a group of beta testers. The app provides information on passenger numbers in real time and also predicts bus and tram usage in the Ginko network. Users can see passenger numbers at any given time as well as forecast usage – by day, line and stop – from data collected by counting systems and predicted peak usage adjusted in real time at each stop. The operator is eagerly awaiting the first results.

In the Netherlands, the city of Twente is using artificial intelligence (AI) to predict transport usage on the basis of weather conditions. With its North Sea coastline, the Netherlands gets mostly mild but also often damp and cloudy weather. This regularly generates extra demand for public transport, notably on weekdays. To optimise its bus service, Keolis Nederland used a deep-learning algorithm to analyse networks in this region bordering Germany. AI matched records of passenger journeys to weather conditions at bus times to reliably forecast passenger numbers. With this information, Keolis Nederland can adapt and optimally configure its services and dispatch buses in accordance with the weather. AI also improved usage forecast accuracy on Twente’s network by 27% during the week and 66% at weekends – a great result!

Reducing overcrowding in public transport, improving safety, enhancing network planning and management and optimising costs – passenger information and usage forecasting systems have become crucial to boost network performance and provide a better passenger experience. No need for a crystal ball to predict that travel demand forecasting solutions have a bright future ahead!