Lighting the way : Platform lighting to enhance train punctuality

• Delays in getting on and off trains seriously impact train departures. By alerting travellers to where the train is least congested, such delays are significantly minimised.
• Carriage CCTV feeds are analysed and utilised in lighting up an illumination strip on the station platform in green, amber or red indicating capacity and carriage and door position.
• Computer Vision number crunching is implemented using small onboard processors and no image data is transmitted off the train.
• Cloud services are used to integrate message feeds on capacity, train position and configuration and publish a lighting activation message to the station display controller.

Making the metro rush less of a crush

This article details how innovative solutions are driving responsive illuminations on train station platforms and are already lighting the way to significantly improved commuter train punctuality.

At busy times, metro trains typically run regularly and at short intervals. The trains are reliable and start their routes on schedule and, yet, often arrive late at stations. The complications of passengers alighting and boarding on crowded platforms cause delays and result in trains starting to backup as they await for slower running services to clear platforms. Platform crowding and customer frustration increases and further delays occur as the situation compounds.

OpenCapacity builds solutions to enable commuters to better visualise the ‘after-alighting-capacity’ of an approaching train, helping them decide where they should optimally position themselves for easier boarding. OpenCapacity’s machine learning predictive modelling of capacity further facilitates future journey planning.

Award winning Innovation

This year, OpenCapacity won the prestigious German Public Transport Mobility Prize — the ‘Deutscher Mobilitätspreis 2018’ for their innovative installation at the Bad Cannstatt, a busy station on Stuttgart ’S-Bahn’ Metro network. OpenCapacity’s installation incorporated a unique visual component: dynamic fibre optic lighting embedded into 200 meters of concrete along the station platform’s edge.

This enabled a realtime routing of passengers through timely guidance illuminations showing exact doors positions in advance and indicating actual carriage capacities. Dynamically illuminated arrows further encouraged commuters to move away from choke points around platform entrances and exits.

Overview of the technical implementation

OpenCapacity installed small onboard computers to capture the carriage CCTV feed and continuously monitor numbers of passengers within each carriage. All computer vision number crunching is done onboard the train. No images are retained and the result is a simple count of the number of people in each carriage area.

On the Stuttgart S-Bahn peak time trains are typically 200 meters in length with 12 carriages. There are 2 different train types (with different door positions) and off-peak trains may vary between 4, 8 or 12 carriages. Every carriage is equipped with CCTV, however, German privacy legislation restricts the images that can be stored or transmitted.

The count for each camera on each train is sent to OC-Cloud where it is stored and used to calculate real time carriage capacity. Also, the large dataset of carriage capacities and other real time feeds is used in a machine learning model that can make predictions of likely train capacity at any date in the future.

When a train is within five minutes of a capacity display enabled station a message is published and the station controller listener starts a display sequence based on capacity and type of train. This display is unique to each station and incorporates features of the station architecture. For example, directing waiting passengers to stay clear of platform stairway exit points.

Predicting capacity ‘after-alighting’ where innovation gets exciting

Useful though it is, displaying the occupancy of an approaching train is not enough. At large transport hubs many passengers will alight from a busy train, making room for others waiting on the platform.

OpenCapacity’s model can predict this ‘after-alighting’ capacity and present it for the approaching train visual notifications and later services. Data feeds on train delays, weather and events are further incorporated to provide more accurate predictions.

The Stuttgart implementation is one example of informing the traveller and trying to influence their behaviour to facilitate both a more efficient service and a more pleasant journey experience. The same data can also be used for operational planning and management to optimise network performance

A further example of predictive modelling can be seen at another OpenCapacity implementation at Shoreditch High Street, London. The Shoreditch implementation uses a large screen display and is also available as a realtime web app at shoreditch.opencapacity.co.