Vassilissa LEHOUX — Naver Labs — Multimodal Mobility by Université Côte d'Azur

Multimodal mobility

Vassilissa Lehoux

8th November 2017

Soph.I.A summit

NAVER LABS

AMBIENT INTELLIGENCE Technology that understands you, contexts, and environment so that you will be given answers and recommendations at the right time without asking

AMBIENT INTELLIGENCE Location Intelligence

Mobility Intelligence

Mapping 3D Hi-def indoor mapping by Robot mapper M1 Web-scale cloud mapping Indoor autonomous mobility

Visual Localization Key technology to find accurate position and orientation with images only

Autonomous Driving Future mobility/traffic data platform Real time road condition/ SAE autonomous level 3

South Korea's best GPS navigation.

Multimodal Mobility

What is multimodal mobility?

Public transportation modes • Bus • Train • Tramway …

What is multimodal mobility?

Public transportation modes • Bus • Train • Tramway … And walking between the stations

Adding some possibilities

Combining • Personal bicycle, car • Taxi • Bike sharing, car sharing … With public transportation

Itineraries in multimodal networks

Door-to-door with • • • •

Single solution for personal modes and transport providers Modes interleaving in one itinerary Proposing relevant options Arriving at a predictable time

Multimodal itineraries Combining different types of modes • • • • • •

Public transportation On-demand public transportation Personnal bike/car, walking Taxi Bike/car sharing …

Into complex mode sequences • Multimodal first and last miles • Multimodal transfers Taxi

Combining modes

Based on Points of Mobility (POM) Bike sharing station

• Stations • Parking lots • etc.

Bike sharing station

Taxi

Combining the different modes Graph based approaches • Aggregation of one graph per mode • Arcs between modes at POM • Public transportation with time-expended or timedependent graphs • Label constrained min-cost paths

Min path methods

Min path methods Taxi first, then public transport (PT) + Walking

Walking

UCCH [Dibbelt, Pajor, Wagner, 2012] SDALT [Kirchler, Liberti, Calvo, 2014] 14

Combining the different modes Time-table based approaches • One graph per mode • Arcs between modes at POM • Public transportation with modeling based on trips/connections

Dynamic programming • Allowing one more trip/connection per round

Combining the different modes Graph based • Methods for road networks are not as efficient on multimodal networks [Bast, 2009] • Large graphs • Automaton might need to be predefined

Time-table based

• Limited precomputations • Sequences can be enforced in the dynamic programming algorithm [Ulloa, Lehoux, Rouland, 2018] Modified RAPTOR with a set of sequences

23 16

Alternative sets Multiple criteria • Earliest arrival time & number of transfers / profile & number of transfers • Min waiting time, min transfer time, min cost, …

Exponential optimal set

A A

Start

Alternative sets Approximation - diversified solutions [Ulloa, 2014] • Fastest route

A • Green line is forbidden

Start

• Violet line is forbidden

Alternative sets Approximation Relevant alternative sets • Constraints and parameters • User preferences and objectives • Diversified sets [Ulloa, Lehoux, Roulland, 2018]

SMIRT Syndicat Mixte Intermodal Régional de Transports

Relevant set sorting • Without personal information • Personalization [Céret, Castellani, Lehoux, 2015] [Wang, Lehoux, Espinouse, Cung, 2018]

Predicting arrival times accurately

Historical data and real-time information

Predicting arrival times accurately

Real-time and historical traffic data • Map matching • Traffic patterns • Planned disturbance • Real-time updates • Short term predictions

Predicting arrival times accurately

Schedule modes information • Time table predictions • Real-time delays • Real-time trip/route modifications

Large scale NAVER LABS • SMIRT (125K stops) • Naver Map (180K stops, 1000K trips)

In the literature • Ile-de-France (110K stops, 290 k trips) • London (21K stops, 133K trips) • Switzerland (25K stops, 1000K trips) • Netherlands (54,5K stops, 620K trips) • Germany (250K stops, 1400K trips)

Large scale NAVER LABS • SMIRT (125K stops) • Naver Map (180K stops, 1000K trips)

Multimodal trip planning Combining all the available modes • Huge networks and many possibilities • Access to transport provider data

Personalization • Giving choices • Respect of privacy

Real-time information • Current state • Short term predictions

Thank you

Bibliography • [Bast, 2009] Hannah Bast. Efficient Algorithms, volume 5760 of Lecture Notes in Computer Science, chapter “Car or Public Transport - Two Worlds”, pages 355-367. Springer, 2009. • [Delling et al., 2012] Daniel Delling, Julian Dibbelt, Thomas Pajor, Dorothea Wagner, and Renato F. Werneck. “Computing and evaluating multimodal journeys”. Technical Report Karlsruhe Reports in Informatics 2012,20, Karlsruher Institut für Technologie (KIT), 2012. • [Dibbelt, Pajor, Wagner, 2012] Julian Dibbelt, Thomas Pajor, and Dorothea Wagner. “User-constrained multi-modal route planning”. In Proceedings of the Fourteenth Workshop on Algorithm Engineering and Experiments (ALENEX), pages 118-129, 2012. • [Dibbelt et al., 2013] Julian Dibbelt, Thomas Pajor, Ben Strasser, and Dorothea Wagner. “Intriguingly simple and fast transit routing”. Experimental Algorithms, Springer Berlin Heidelberg, Berlin, Heidelberg, pages 43-54, 2013.

Bibliography • [Kirchler, Liberti, Calvo, 2014] Dominik Kirchler, Leo Liberti, and Roberto Woler Calvo. “Efficient Computation of Shortest Paths in Time-Dependent MultiModal Networks“. Journal of Experimental Algorithmics (JEA), Volume 19, 2014, Article No. 2.5. • [Strasser and Wagner, 2014] Ben Strasser and Dorothea Wagner. "Connection scan accelerated”. Proceedings of the Meeting on Algorithm Engineering & Expermiments (ALENEX’14), pages 125-137, Portland, Oregon, 2014. • [Witt, 2015] Sascha Witt. “Trip-Based Public Transit Routing”. In: Bansal N., Finocchi I. (eds) Algorithms - ESA 2015. Lecture Notes in Computer Science, vol 9294. Springer, Berlin, Heidelberg. • [Céret, Castellani, Lehoux, 2015] Eric Céret, Stefania Castellani and Vassilissa Lehoux. “System and method for multi-factored-based ranking of trips”. Patent application US20170053209A1, Xerox Corp. • [Bast et al. 2016] Hannah Bast, Matthias Hertel and Sabine Storandt. “Scalable Transfer Patterns”. In Proceedings of the Eighteenth Workshop on Algorithm Engineering and Experiments (ALENEX), 2016. © 2018 NAVER LABS. All rights reserved.

Bibliography • [Delling et al. 2017] Daniel Delling, Julian Dibbelt, Thomas Pajor, and Tobias Zündorf. “ Faster Transit Routing by Hyper Partitioning”. Proceedings of the 17th Workshop on Algorithmic Approaches for Transportation Modelling, Optimization, and Systems (ATMOS 2017), article No. 8; pages 8:1–8:14, Open Access Series in Informatics. • [Ulloa, Lehoux, Roulland, 2018] Luis Ulloa, Vassilissa Lehoux-Lebacque and Frédéric Roulland. “Trip planning within a multimodal urban mobility”. IET Intelligent Transport Systems, Volume 12, Issue 2, pages 87 –92. • [Wang, Lehoux, Espinouse, Cung, 2018] Lizhi Wang, Vassilissa Lehoux, Marie-Laure Espinouse and Van-Dat Cung. “Multimodal itineraries ranking using fuzzy logic”, Operations Research 2018 conference (OR2018), Brussels, Belgium, 2018.

SMIRT’s mobility companion plateform

SMIRT’s Request for Proposal (France Nord Region) • Integrates all urban transport service providers information • 14 transportation authorities Main features • Multimodal passenger information / trip planning • Real-time transit information • Personalized trips

Multimodal trip planning

Example of complex mode sequences for SMIRT

Mode 1 Anywhere

POM Type 1

POM Type 4

Mode 2

Scheduled mode

Mode 3 POM Type 2

POM Mode 5 POM Type 5 Type 6

Scheduled modes

Mode 6

POM Type 7

Mode 3

Mode 7

POM Type 3

POM Type 7

Mode 4 Anywhere

Scheduled mode

POM Type 8

Multimodal trip planning

Example of complex mode sequences for SMIRT

Start

Bike sharing station

Bus Stop

Bike sharing station

Bus Stop

Scheduled modes

Bike sharing station

Taxi Station

Tramway Stop

End

Tramway Stop