Modelling 2: TRAFFIC FORECAST MODELS
Leaders:
WP3: Igor Grabec (Amanova), Pär Ekström (Caran)
Description:
Traffic forecast models.
WP2 question:
To proceed with WP2 task 2.2 "New types of data needs" please determine the data requirements for your 'idea' untill August, 22!
Pekka Leviäkangas summarises the recent discussions on traffic forecast models in ROADIDEA as follows:
Models - traffic, weather, road condistions, pavements,
bridges, etc. - produce a lot of different types of information and so far we
have mainly listed data sources by considering data bases or data banks.
However, these models run by road administrations or affiliated organisations
produce information that in a way could be regarded as public information. The
information from models could well serve as a basis for new innovative
services. Hence, the model discussion was very relevant in ROADIDEA context.
Basically, we would need a similar analysis, as we did for data bases and
real-time sensor networks, regarding various models. This is probably out of the
scope in such extent, but could still be considered in a number of places in
this project.
I think it was also concluded in the discussion that
ROADIDEA will not develop new models for traffic forecasting or weather
forecasting or such, but it could well contribute to existing models by
providing innovative additional components, sub-models, value-adding
integrated models, etc. This falls natural within the scope of
ROADIDEA. WP2, WP3, and WP4 could all contribute to this. I see the
greatest role for WP3, which is about methods and modelling.
The current Task Allocation Plan (the excel sheet with
more detailed task descriptions and partner roles) does not explicitly exclude
this possibility, but neither does it support it directly. But good ideas should
not be suppressed by project bureaucracy. Particularly so in this
project.
Marcus Wigan observes that:
While we will not develop new models for traffic flow(we do not have the resources and there is a good choice out there already) we DO have to take advantage of available information to synthesise specific user services.. and this may be done both from a combination of sensor inputs and other data feeds- and also by using traffic flow models (maybe short run time series kinds) as part of the data fusion process.
Bayesian updating is one such technique and is very much in the RoadIdea frame. As RoadIdea is concerned with user services- and thus fairly short term estimation horizons- the more sophisticated data fusion approaches exemplified by thsoe described(and downloadable from) www.opus-project.org are probably also excluded as they run rather slowly.. but- if anyone wishes to examine these methods, the entire source code in in Open Source and available.. if it is not all on the site specifed, I can arrange it be...
The data flows and sources for traffic data are beginning to become standardised in many countries. the ADUS committee of the US TRb has produced both a US ASTM Standard on ITS data, and also is monitoring the practical results of efforts to work to it (there is a session at TRB2009 on this alone). Here are links for the current activity.
Quality Control Procedures for Archived Operations Traffic Data: Synthesis of Practice and Recommendations
<http://www.fhwa.dot.gov/policy/ohpi/reports/qcproc/index.htm>www.fhwa.dot.gov/policy/ohpi/reports/qcproc/index.htm
The publication has been posted and listed on the ADUS page:
<http://www.fhwa.dot.gov/policy/ohpi/travel/adus.htm>www.fhwa.dot.gov/policy/ohpi/travel/adus.htm
Drawn from the ADUS committee circulations(im on ADUS)
The role of extending Standars (WP29, TUV,ASTMS, ISO...) offers notable leverage of anything we come up with.,.