1 Preferred citation style for this presentation Axhausen, K.W. (2006) Next steps ?, MATSIM-T...
-
Upload
rosamund-white -
Category
Documents
-
view
212 -
download
0
Transcript of 1 Preferred citation style for this presentation Axhausen, K.W. (2006) Next steps ?, MATSIM-T...
1
Preferred citation style for this presentation
Axhausen, K.W. (2006) Next steps ?, MATSIM-T Developer Workshop, Castasegna, October 2006.
Next steps ?
KW Axhausen with comments by K Nagel
IVTETHZürich
October 2006
3
Conceptual starting point
Competition for slots on networks and in facilities
Activity scheduling
k(t,r,j)i,n
qi ≡ (t,r,j)i,n
Mental map
4
Upcoming projects and possibilities at ETH
• KTI – Speeding it up (7 PY)• SNF – Influenca dissemination (1/2 PY)
• NFP 54 – Parking supply agent (decision by October)• ETH – Retail supply agent (decision by March)
• VW – Social networks & travel (second stage in preparation)
• BAK – Road pricing application (first talks)• Kanton – Uetlibergtunnel impact (first talks)
5
Upcoming projects and opportunities at TU Berlin
• VOLVO (2PM left) -- dynamic road pricing
• COOPERS (10 PM) -- telematics scenarios
• Tsunami evacuation (3PJ) -- application pending
• VW – Social networks & travel (second stage in preparation) (see above)
6
Other initiatives
• PTV project for an XML-kernel for data exchange between models (BMBV)
• PUMA, TU/E, CEMDAP, PB and CS in search of a DTA ?
7
What do we have to do in the next 12 months ?
• Interface to TeleAtlas and to commerical databases
• Overnight run for 8 mio agents to steady state (with small parallel machine ?)
• Mode choice (planomat or brain ?)• Routers for public transport, cyclists, pedestrians• Better chains, durations and timings• Destination choice at parcel level
• Parameter estimation for planomat• Validation phase
8
What do we have to do in the next 12 months ?
• Stabilize matsim on sourceforge• Regression testing• Moving towards better code robustness
• User interface?
• (Output database)• Analysis tools
9
First ideas for the to do list
Mode choice (in the steady state context):
• Travel time surface for any x,y – pair and any time t for all modes (intermodal routes ?) (taste differences ?)
• Cost model parking; tolls• Cost model public transport
Speed: • Better initialisation of speeds (speed surfaces)• Reuse of travel time surfaces between scenarios• More intelligent replanning rates
• Scheduling module (GA, CMA, Ant optimisation, etc. ?)
10
First ideas for the the to do list
Better chains as inputs:
• Larger range of chains (realistic trip rates)• Coding of chains also by duration class
• Improved (variation of) time windows
Validation:
• Quality of population synthesis• Traffic counts (and their comparability)• Use of the speed estimates from the FCD data
11
Other goals
• Plug & play
• Quasi – commercialisation
• Behavioural richness
• Steady state and evolution
12
Plug & play ?
Do we want to invest into
• XML – standards (e.g. RailML ?)
• Integration into OPUS • Interface to PUMA
• Interface to VISUM/VISSIM
• Interfaces to Tasha, Albatross, FAMOS, CEMDAP
13
Quasi-commercialisation
Target: • Consultancy or research group with sound GIS/data
management capabilities• Some programming skills
Requirements: • Data conversion and imputation tools• Integration with BIOGEME and parameter estimation tools• Run management and data extraction/comparison• Visualisation and animation (Link to OpenEV or Saga)
• Better documentation, starter kit, www.matsim.org
14
Behavioural richness
• Agenda generation (Habib & Miller)
• Household interaction• Shared trips• Timetable coordination
• Social network interactions • Location choice• Timetable coordination
• Capacity constraints at destinations
15
Behavioural richness
• Intermodal routes
• Richer traffic flow model
• Traffic control systems• Parking and parking search
• Supply side agents• Parking• DRT• Shopping
16
Evolution over multiple days (periods)
• Dynamic agenda generation• Seasonality
• Communication between agents
• Generating the initial mental map• Storing the mental map• Updating the mental map
17
Priorities
• Share of effort for development and for interfaces ?
• What do we want to do when ?
• Who does what when ?
• How do we choose ?• How do we monitor progress ?