jan$lundgren$...
TRANSCRIPT
The Swedish Postgraduate School in Intelligent Transport Systems
Jan Lundgren ITS World Congress, October 8, 2015
The Swedish Postgraduate School in Intelligent Transport Systems
CooperaDon between ITS Sweden and seven universiDes
Supported by the Swedish Transport AdministraDon and the Swedish Governmental Agency for InnovaDon Systems (VINNOVA)
Aim:
• Improve the PhD-‐educaDon through specialised doctoral courses and extended supervision of students
• Promote academic research within the field of ITS relevant for industry and society
• Increase the interest and focus of transportaDon research and educaDon within the universiDes
• Establish networks for collaboraDon both naDonally and internaDonally
The Swedish Postgraduate School in Intelligent Transport Systems
Modeling and Simula0on Study of Heavy-‐Duty Vehicle Platooning
Qichen Deng
KTH Royal InsDtute of Technology
Research ques0ons
• What are the benefits of platooning heavy-‐duty vehicles on highway (HDVs)?
• When can two HDVs form a platoon?
Benefits – Improve Fuel Efficiency of HDV
Benefits – Improve Traffic Efficiency
0 50 100 1500
500
1000
1500
Traffic Density [veh/lane/km]Traf
fic F
low
Rat
e [v
eh/la
ne/h
our]
No HDV PlatooningHDV Platooning with CVS Policy
0 50 100 1500
500
1000
1500
Traffic Density [veh/lane/km]Traf
fic F
low
Rat
e [v
eh/la
ne/h
our]
No HDV PlatooningHDV Platooning with CVS Policy
0 50 100 1500
500
1000
1500
Traffic Density [veh/lane/km]Traf
fic F
low
Rat
e [v
eh/la
ne/h
our]
No HDV PlatooningHDV Platooning with CVS Policy
0 50 100 1500
500
1000
1500
Traffic Density [veh/lane/km]Traf
fic F
low
Rat
e [v
eh/la
ne/h
our]
No HDV PlatooningHDV Platooning with CVS Policy
(a) 10% of total traffic is HDVs; (b) 15% of total traffic is HDVs; (c) 20% of total traffic is HDVs; (d) 25% of total traffic is HDVs.
(a) (b)
(c) (d)
When can two HDVs form a platoon?
• Key factors affecDng the HDV platooning formaDon:
1. Traffic density
2. Driving behavior of passenger car
3. Speed of HDVs
Lane 1
Lane 2
Platoon Forma0on of Two HDVs on a Two-‐Lane Highway
16 18 20 22 24 26 28400
600
800
1000
1200
1400
1600
1800
2000
Number of Vehicles between Two HDVS plus Traffic Density of Lane 2
HD
V P
lato
on F
orm
atio
n Ti
me [
s]
ReferenceFront HDV Speed 70km/hFront HDV Speed 75km/hFront HDV Speed 80km/h
16 vehcles1080s
18 vehicles540s
17 vehicles720s
Coopera0ve ITS in traffic management
Ellen Grumert Linköping University
Coopera0ve ITS in traffic management -‐ Mo0va0on
Source: Foto taken by user Fir0002, publiced at www.wikipedia.org (accessed 2014-‐09-‐21)
𝒒< 𝒒↓𝒄
Density
Flow
Coopera0ve ITS in traffic management -‐ Approach and method
InformaDon
InformaDon
InformaDon
InformaDon
InformaDon från central styrning
Source: Foto taken 2010 by Holger Ellgaard, publiced at www.wikipedia.org (accessed 2011-‐04-‐13)
V2V
I2V/V2I
SUMO
Coopera0ve ITS in traffic management -‐ Main conclusions
• The performance of the variable speed limit systems is dependent on the choice of algorithm for deciding on which variable speed limit to use on the road.
• CooperaDve variable speed limit systems using I2V harmonize the flow compared to regular variable speed limit systems.
• LimitaDons in having fixed measurement points. V2V could probably improve the performance further allowing for measurement point in between detectors.
• Speed limits reflecDng the condiDons on the road gives best performance.
• Early prevision/early detecDon limits the effects of high flows.
• The capacity levels are of high importance for the performance.
Performance evalua0on of coopera0ve awareness in C-‐ITS
Nikita Lyamin
Halmstad University
Background
Research problems
Dynamic Origin-‐Des0na0on Matrix Es0ma0on for off-‐line applica0ons
Athina Tympakianaki
KTH Royal InsDtute of Technology
OD es0ma0on problem
• OD estimation is important in many applications - essential input traffic simulation models (microscopic, planning models)
- traffic management (ITS systems)
- traffic prediction - transportation planning
- evaluation of different strategies
• Find an OD matrix that best matches a set of indirect observations (e.g. counts, speeds)
• Underdetermined problem, many unknowns fewer equations: different OD matrices result in the same traffic counts
• Addition of more data sources reduces the set of possible solutions – requires more general formulations of the problem and solution algorithms
• Traffic simulation is used to capture the traffic conditions resulting from different OD matrices
Proposed OD es0ma0on methods 1. A gradient algorithm with formulation based on an assignment matrix that maps OD flows to counts at sensor locations (e.g. Toledo and Kolechkina, 2013). 2. Cluster-based SPSA algorithm (Tympakianaki, Koutsopoulos, Jenelius, 2014) - Modification of the commonly used SPSA (Spall, 1998) algorithm. Motivation: • Different OD magnitudes • Unstable performance, algorithm diverged
Modified algorithm, cluster-based SPSA: • OD pairs are clustered based on some criteria:
§ Homogeneity within cluster § Magnitude of OD flows § Size of cluster § Number of clusters
• Gradient is approximated independently for each cluster
Case study – synthe0c data • Södermalm network, Stockholm • 1100 urban and freeway and urban links • ’True’ OD demand: synthetic data
- 462 OD pairs - 7-8 a.m. study period - 15 min time interval
• Sensor coverage: 5% of the total number of links. • Traffic observations: counts, speeds • Mesoscopic traffic simulation model: Mezzo
0
100000
200000
300000
400000
500000
600000
0 200 400 600 800 1000 1200
Objec0v
e func0o
n value
Number of func0on evalua0ons SPSA c-‐SPSA
0
10
20
30
40
50
60
70
IniDal error SPSA c-‐SPSA RM
SE
OD flows (veh/h) Counts (veh) Speeds (km/h)
• More robust algorithmic performance • More accurate OD matrices • Less sensitive to the selection of the algorithmic parameters • Improved practical convergence
Motorway applica0on using real data
How well does the method esDmates the OD flows when an incident occurs?
OD esDmaDon: • Gradient descent algorithm • MCS counts Counts (veh/h)
IniDal error Final error
RMSE 103 56
ME 29,3 0,83 -‐100
-‐50
0
50
100
150
200
250
6:00 6:15 6:30 6:45 7:00
OD flo
w cha
nge (%
)
Time (hh:mm)
OD pair 2 OD pair 3 OD pair 1
On-‐ramp
Off-‐ramp
Incident
Detour
Södertälje -‐ Bredäng
14 km 14 OD pairs
Transport for sustainable urban development – integrated modelling of
walk, cycle and public transport
Gerasimos Loutos Linköping University
Project View
The goal is to improve the modeling of bicycle trips in current demand models
AIM to: • Develop techniques for modeling trips chains including bicycle and public transport legs.
• IdenDficaDon of appropriate level of detail for non-‐motorized modes. • IdenDfy factors important for bicycle mode choice and factors important for policy analyses.
Policy variables – Factors that influence the share of cyclists
Built environment Urban form (distance), infrastructure and faciliDes at work
Natural environment Hilliness, climate and weather
Socio-‐economics Gender, age, income, car and bicycle ownership, employment status
Psychological factors Amtudes, social norms and habits
Cost, travel 0me, effort & safety
Mode choice models
Built environment Urban form (distance), infrastructure and faciliDes at work
Natural environment Hilliness, climate and weather
Socio-‐economics Gender, age, income, car and bicycle ownership, employment status
Psychological factors Amtudes, social norms and habits
Cost, travel 0me, effort & safety
Mode choice models es0ma0on
Survey Data (Revealed Preference, ex. RVU) • InformaDon on conducted trips & socio-‐economic variables • Drawback: only characterisDcs for individual’s chosen mode
I. e. Missing travel Dme and distance (non-‐used) alternaDves
Regression Models Trip planning applikaDon
Available tools and future development Currently used models
• Sweden: SAMPERS, Local EMME or VISUM-‐models, SLL VISUM-‐model • Do not include specific bicycle and walk factors (e.g. infra, costs) • Cycity: project targets to improved data collecDon (more surveys, GPS data
from cyclists etc.) • MoveMeter (MoveMobility), Brutus (Strafica) in a Swedish semng.
Frame a new transport demand tool • Include modelling trips chains with several modes • Assignment of trips to the street (OpenStreetMap) network for all modes using,
• Time table data for public transport for walk/PT-‐trips • Bicycle rouDng using a trip planner (in OSM network) • TradiDonal car assignment (in OSM network)
Access management in intermodal freight
transporta0on
Stefan P G Jacobsson Chalmers University
Access management in intermodal freight transporta0on -‐ Iden0fied problems1
Decentralised
Many actors Different modes
Deficiencies in the interaction
Practical problems in the hubs
Lack of high-quality real-time data2
Decentralised
Decentralised
Decentralised Decentralised
________________________ 1 According to Marchet et al., 2012, Perego et al., 2011, Bujis and Wortmann 2014 2 According to SteadiSeifi et al., 2014
Access management for carrier operaDons in intermodal freight transportaDon
Research gap Research agenda /
Next steps
(1) Problem idenDficaDon and
moDvaDon
Con-‐versaDons
Obser-‐vaDons
Literature study
Bench-‐marking
(2) Define the objecDves for a soluDon
Access management
Handling of carrier access
Beper understanding and knowledge
Literature study
Con-‐versaDons
Obser-‐vaDons
Access management in intermodal freight transporta0on -‐Methodology
• For smaller actors • Use already exisDng IT-‐systems • InteracDon in real-‐Dme • Qualify for access to an intermodal container
terminal
Access management in intermodal freight transporta0on -‐ Solu0on
Development & Evalua0on of Mobile Phone App for Detec0ng Safety Cri0cal Event of Motorcycle Rider
Noor Azreena Kamaluddin
Lund University
Introduc0on
q An app for motorcyclists will be developed to register safety critical
events. q Safety critical events will be identified by certain trigger
variables (e.g. jerk). q The app helps to describe the course of events in a (near)
accident. q The effects and rider acceptance of the system will be
studied in a real life study.
Background
Ø Safety : crucial issue of motorcycle (More than 60% of fatalities, 20x higher risk)
Ø Shortfalls issue – Complementary method (Reliable estimation)
Ø Vital prerequisite to set-up more realistic targets for crash/casualty reduction programmes
Ø No available or standard procedure in use for collecting self-reported accidents in Malaysia
Objec0ves
Ø Collect self-reported accidents & determine under-reporting - GAP Ø Develop & test of a mobile apps (e-Call) for motorcyclists – enrich current
system Ø Identify/predict safety critical events based on sudden acceleration
change (jerk) Ø Produce complementary method of accident recording system Ø Map locations with a higher single-accident risk based on the smartphone
application (eCall app)
The eCall System for Motorcyclist
Usefulness: Ø Send/prepare message to
SOS centre
Ø Minimize rate of fatalities: Immediate medical assistant
Ø Collection of anonymous data & identify bike-accident prone area: Good basis of prevention