Intermodality is the combination of several modes of transport in one trip. From an individual mode to a collective or public mode (public transport, bicycle, etc.). It should be distinguished from multimodality, which describes the existence of several modes to connect two places. Faced with congestion generated by car use or increasing pressure on infrastructure, intermodality and multimodality are among the primary objectives of the Mobility Authorities and transport operators.
Access to reliable and predictive information in real time makes it possible to better guide users’ mobility choices. For this purpose, Qucit develops predictive motors to know in advance the availability of self-service bicycle systems (station by station), the filling of the car parks or the time needed to find a parking space on the road. This promotes intermodality by providing reliable information to the user when planning his route.
An opinion shared by Île-de-France Mobilités, the Paris region Transport Authority. For Jean Luc Prat, in charge of multi-modal information: “The future of transport systems will be precisely’ multi-modal’. The combination of transport offers, associated ticketing services and the optimization of passenger information will enable the development of a real-time and predictive mobility assistant to facilitate the traveler’s journey.”
BikePredict: Predicting the availability of bike-share
Bike-share systems are booming in cities. More than a thousand cities in the world have one and the Vélib in Paris is one of the largest.
Their practical and economical qualities make cycling a reliable transport mode. It is particularly useful for last mile connectivity and short distances.
To support urban cycling, the Île-de-France Mobilités transport Authority is working on ViaNavigo, a website and mobile application which provides quality information on transport in the Paris region, particularly on self-service bicycles.
Last year, Île-de-France Mobilités integrated in ViaNavigo the BikePredict API (Application Programmable Interface – a software package that is easy to integrate into a website or mobile application).
Thanks to the API developed by Qucit and GeoVélo’s itinerary profiles, Vélib is a public transport like the others. Anyone can use the Vianavigo route planner from Île-de-France Mobilités to choose a route that combines metro and Vélib’ for example. And most importantly, you are sure to find a bike when you go out on the street and a spot to drop it off near your final destination.
And it’s a success! Jean Luc Prat adds:“Thanks to predictive prediction, Vianavigo can respond with precision to the users’ route research by enhancing the service provided. We started with the Bike sharing mode, it is now necessary to be able to extend predictive algorithms for public transport and eventually on car journeys”.
ParkPredict: Predicting Park&Ride car parks availability
The growth of Park & Ride car parks is not a new phenomenon. Today, they are a privileged means of reducing the use of private cars in dense city centers and ensuring intermodality and the use of public transport. A Park & Ride allows motorists to leave their private car on the outskirts of the city and end their journey by public transport with stations or stops.
One of the key factors for intermodality and in particular for the use of parks & rides is information. In addition to proper signage during the journey, a motorist will use this option if he or she is sure to find a P+R spot.
This is why we work on artificial intelligence models that predict the availability of parks & rides and work with partners such as Keolis Bordeaux Métropole to provide this valuable information in real time to motorists wishing to travel to city centers.
The Bordeaux example
With the arrival of the tramway in the metropolitan territory in the early 2000s, a large number of Park & RIde car park helped Bordeaux Métropole’s efforts to limit car use in the city center and promote intermodality. 5506 spots are available for motorists wishing to get to the Bordeaux center. Two additional P+Rs are being delivered. The majority of the 23 parks & rides are nowadays victims of their success. It is necessary to inform motorists in advance about the availability of these parks. The predictions will be integrated into the Infotbm.com website and mobile application by the end of the first quarter of 2018.
How does it work?
The algorithm behind BikePredict & ParkPredict is a prediction machine. Every minute, the software developed by Qucit analyzes availability data, and a multitude of variables to understand what is happening in the city. Thus, we can predict the number of bicycles available in each station, or the number of spots available in each station, or the availability of Park & Ride parks.
In order to calculate bike and parking predictions, we use machine learning models (Gradient Boosted Trees, Quantile Regression, Distributed Asynchronous Hyperparameter Optimization). Those models are fed with historical and contextual data.
In addition to the activity data from relay parks or self-service bicycle stations, various variables are added depending on the context:
- The calendar (days of the week, holidays…)
- Weather (precipitation, wind, temperature, visibility, cloud cover…).
- Real-time traffic around P+Rs and in the city
- OpenStreetMap data (networks and stops, roads, public facilities…).
In the learning phase, these contextual data allow models to understand the variables that impact on real-time and future parking demand in car parks and bike sharing stations. These models are then able to reliably predict this demand at different time horizons (15 minutes to 24 hours) from the real-time context.
Île-de-France Mobilités (formerly STIF) is the organizing authority for transport in Île-de-France. A key player in the network, it organizes, decides, invests and innovates to improve mobility and services to travelers.
Île-de-France Mobilités has 4.3 million regular navigation users in 2016.
Île-de-France Mobilités and Qucit have been working together since 2017 to provide predictive information on the Vélib system to Ile-de-France residents via the ViaNavigo application.
Keolis Bordeaux Métropole
Keolis Bordeaux Métropole, a subsidiary of the Keolis Group, has been in charge of operating the TBM network since 2008.
The TBM network, which serves the 28 cities of Bordeaux Métropole, consists of 3 tram lines, 78 regular bus lines, 2 BatCub lines (river shuttle service) and 1800 VCub, self-service bicycles plus 23 Park&Rides.
Keolis Bordeaux Métropole and Qucit have been working together since 2014 to provide a more fluid global mobility solution for residents of the Bordeaux metropolis:
- Keolis Bordeaux Metropole has integrated in 2014 a predictive bike sharing solution (BikePredict) in its modal application dedicated to cyclists (La Bonne Station).
- Keolis Bordeaux Metropole has integrated in 2016 a predictive time solution for parking on-street (ParkPredict On Street) in its travel calculator available on its website. Predictive information displays information on the total time it takes to travel by car from point A to point B.