MATSim developments in France
and at IRT SystemX

Sebastian Hörl

17 February 2021

eqasim

Population synthesis

Travel demand synthesis

Mobility simulation

(MATSim)

  • Framework to easily set up runnable MATSim simulations
     
  • Package of tested and calibrated out-of-the-box runnable and customizable simulations
     
  • Reproducibility from raw input data to final analysis

Data

Analysis

eqasim

Population synthesis

Travel demand synthesis

Mobility simulation

(MATSim)

Data

Analysis

Discrete choice models

+

  • See ABMTRANS 2021

Census data

Récensement de la population

RP

Icons on this and following slides: https://fontawesome.com // Background: Simunto VIA

Île-de-France

Dispositif sur les revenus localisés sociaux et fiscaux
Income tax data

RP

FiLo
SoFi

 

Île-de-France

RP: Flux de mobilité
Commuting data

RP

RP
Mob

 

FiLoSoFi

Île-de-France

Enquête globale de transport
Household Travel Survey

RP

EGT

FiLoSoFi

RP Mob

Enquête national transports et deplacements
Household Travel Survey

ENTD

Île-de-France

Enquête globale de transport
Household Travel Survey

RP

EGT

FiLoSoFi

RP Mob

Enquête national transports et deplacements
Household Travel Survey

ENTD

Île-de-France

SIRENE
Enterprise census

RP

S

FiLoSoFi

RP Mob

EGT

ENTD

BD-TOPO
Address database

BD

Île-de-France

RP

FiLoSoFi

RP Mob

EGT

ENTD

SIRENE

Person ID    

Age

Gender

Home (X,Y)

1

43

male

(65345, ...)

2

24

female

(65345, ...)

3

9

female

(65345, ...)

BD-TOPO

Île-de-France

RP

FiLoSoFi

RP Mob

EGT

ENTD

SIRENE

Person ID    

Activity

Start

End

Loc.

523

home

08:00

(x,y)

523

work

08:55

18:12

(x,y)

523

shop

19:10

19:25

(x,y)

523

home

19:40

(x,y)

Person ID    

Mode

Start

End

523

Public T.

08:55

523

Public T.

18:12

19:10

523

Walking

19:25

19:40

08:00

BD-TOPO

Île-de-France

OpenStreetMap
Road network

RP

OSM

FiLoSoFi

RP Mob

EGT

ENTD

IDFm GTFS
Public transport schedule

GTFS

SIRENE

OSM

GTFS

Île-de-France

BD-TOPO

Open
Data

Open
Software

+

=

Reproducible research

Verifiable results

Integrated testing

  • Hörl, S. and M. Balac (2021) Open data travel demand synthesis for agent-based transport simulation: A case study of Paris and Île-de-France, Under review.

RP

FiLoSoFi

RP Mob

EGT

ENTD

SIRENE

OSM

GTFS

Île-de-France

BD-TOPO

Even more data!

  • Since January 2021 opening up of most of the data of the National Institute of Geography
  • Detailed road network for all France
  • Detailed building footprints with height, utilisation, number of appartments, ...

RP

FiLoSoFi

RP Mob

EGT

ENTD

SIRENE

OSM

GTFS

Other regions ...

BD-TOPO

Replace for Lyon, Toulouse, Lille, ...

Nantes (Université Gustave Eiffel)

  • Population synthesis
  • Noise modelling
     
  • Contact: Valentin Le Bescond

Current use cases

Lille (IMT Lille Douai)

  • Park & ride applications
  • Road pricing
  • See ABMTRANS 2021
     
  • Contact: Azise Diallo

Current use cases

Toulouse (Odyssee)

  • Placement and use of shared offices
     
  • Contact: Vincent Loubière

Current use cases

Rennes (Tellae)

  • Micromobility services
  • Custom simulator: Starling
     
  • Contact: Vincent Leblond

Current use cases

Paris / Île-de-France
(LVMT / Ecole de Ponts ParisTech)

  • Scenario development for sustainable urban transformation
     
  • Contact: Nicolas Coulombel

Current use cases

Paris / Île-de-France
(Université Gustave Eiffel / GRETTIA)

  • Scenario development for areas in Paris
     
  • Contact: Mahdi Zargayouna

Current use cases

Paris / Île-de-France
(IFP energies nouvelles)

  • Simulation of cycling and connection to crowdsourced data (see ABMTRANS 2021)
     
  • Simulation of home office used based on Covid surveys
     
  • Contact: Alexandre Chasse

Current use cases

Paris / Île-de-France
(IRT SystemX)

  • Simulation of automated shuttle services (see later today)

Current use cases

Lyon (IRT SystemX)

  • Low-emission parcel delivery simulations

Current use cases

Current use cases

Getting started

Documentation where to get the data sets

git clone https://github.com/eqasim-org/ile-de-france
conda env create -f environment.yml
python3 -m synpp config_ile_de_france.yml

Getting started

Documentation where to get the data sets

git clone https://github.com/eqasim-org/ile-de-france
conda env create -f environment.yml
python3 -m synpp config_ile_de_france.yml

Getting started

Documentation where to get the data sets

git clone https://github.com/eqasim-org/ile-de-france
conda env create -f environment.yml
python3 -m synpp config_ile_de_france.yml

Getting started

Documentation where to get the data sets

git clone https://github.com/eqasim-org/ile-de-france
conda env create -f environment.yml
python3 -m synpp config_ile_de_france.yml

Also ...

LEAD

  • "Digital twins for low emissions last-mile logistics"
     
  • Use case: Confluence area in Lyon
     
  • Scenario development for parcel delivery
    • Population / business growth
    • Use of an urban distribution center with cargo-bikes, delivery robots, ...

Project idea

Lyon

Confluence

LEAD

First proof-of-concept

Baseline

  • Synthetic population
  • Generated deliveries
  • Background freight traffic

Hub analysis

Case scenarios

Hub analysis

Future scenarios

  • Sizing of distribution center
     
  • Sizing of distribution fleet
  • Size of distribution vehicles
     
  • Shared of deliveries
    (Strict / soft zero-emission zone)
     
  • Which configurations are feasible and effective?
  • Increase of population
  • Change of demographics
  • ... and generated deliveries
     
  • Static increase of background freight (Interface Transport)

+

LEAD

Study design

  • Framework to run MATSim simulation inside of calibration loop



     
  • Multiple approaches available:
    • SPSA, CMA-ES, basic Opdyts, ...
       
  • Internal research project started on:
    • Calibration obectives
    • Convergence metrics
    • Hyper-parameters

LEAD

  • Calibration of choice model parameters using SPSA

Calibration

  • Calibration of choice model parameters using SPSA

Calibration

  • Calibration of choice model parameters using SPSA

Calibration

  • Convergence criterion is crucial for calibration!
     
  • Here: Share of persons changing any trip in the plan after applying the choice model must be lower than 5% of the population (+ see ABMTRANS 2021)
     
  • Fixed stopping criterion will likely break calibration:

     

Hörl, S., Becker, F., Axhausen, K.W. (2021) Simulation of price, customer behaviour and system impact for a cost-covering automated taxi system in Zurich, Transportation Research Part C: Emerging Technologies, 123, 102974.

Calibration

  • Calibration of counts?                Beware of stuck agents!
  • Joint calibration seems necessary
     

  • Currently experimenting with calibration of car parameters and capacities by OSM road type and count objective (no mode share objective yet).

Calibration

  • Calibration of counts?                Beware of stuck agents!
  • Joint calibration seems necessary
     

  • Currently experimenting with calibration of car parameters and capacities by OSM road type and count objective (no mode share objective yet).

Calibration

  • Research project with MOIA
  • Increasing compatibility of DRT and AMoDeus



     
  • Currently 2020w45, but works with latest Open Berlin

Developments around DRT

Developments around DRT

  • Equipping the pom.xml ...
<repository>
	<id>amodeus-mvn-repo</id>
	<url>https://raw.github.com/amodeus-science/amodeus/mvn-repo/</url>
</repository>

[...]

<dependency>
	<groupId>amodeus</groupId>
	<artifactId>amodeus</artifactId>
	<version>2.1.0</version>
</dependency>

Developments around DRT

  • Equipping the code ...
AmodeusConfigGroup amodeusConfig = new AmodeusConfigGroup();
controler.getConfig().addModule(amodeusConfig);
		
AmodeusModeConfig modeConfig = new AmodeusModeConfig("drt");
amodeusConfig.addMode(modeConfig);
		
modeConfig.getDispatcherConfig().setType("HighCapacityDispatcher");
//modeConfig.getDispatcherConfig().setType("TShareDispatcher");

controler.addOverridingModule(new AmodeusDrtModule());
AmodeusDrtModule.overrideDispatchers(controler, controler.getConfig());

Developments around DRT

  • Current features
    • Run dispatchers such as Alonso-Mora or TShare
    • Take into account detour / wait time constraints from DRT
    • Rejections as in DRT
    • Optionally, use DRT relocation algorithms
       
  • Reimplementation of Alonso-Mora algorithm
    • Apparent similarities with DRT algorithm
       
  • Future
    • Further thin out MATSim parts in AMoDeus to have library of algorithms
    • or Move algorithms to MATSim (follow relocation)

Something that may become useful ...

  • Strong point of AMoDeus is/was visualisation when assessing the functioning of different algorithms
     
  • Unfortunately, no compatibility with DRT / MATSim
     
  • I picked up some old experimental code to see better what is going on

Ongoing thesis

  • Tarek Chouaki: Application of Reinforcement Learning

Thank you!

Contact: sebastian.horl@irt-systemx.fr

Seminar TUB

By Sebastian Hörl

Seminar TUB

Zoom, 17 Feb 2021

  • 306
Loading comments...

More from Sebastian Hörl