TheCity

Hi Don,

I am posting this message against the original in the
http://thecity.sourceforge.net forum as well so others can find it
plus in my related simulation:

http://trafix.sourceforge.net

The code uses a second site I established for C++ allocator
development (See Bjarne Stroustrup, 3rd Ed, Sec 19.4). The trafix
code is dependent on the allocator available from:

http://allocator.sourceforge.net

I am under the impression that the freeware traffic simulation action
is rudimentary. I was working for a time at the US Department of
Transportation's Volpe Research Center in Cambridge, and I sent a memo
to the director of the center suggesting hosting an open source
traffic simulation website, but at the time the idea was ruled out as
competition with their for-profit traffic simulation venture with MIT.
"Of the people, by the people, for the people"... only goes so far.

One of the interesting issues with an overall traffic/city simulation
umbrella is the programming language issue. If each simulation uses a
different language, does that pose a problem? Might it be better to
create a site which breaks out simulations based on language
implementation, and then at a higher level, have a related section
which describes traffic algorithms and models. This latter algorithm
section would not be language dependent.

Not having full access to SimCity I think would be a problem. One of
the traffic simulation aspects I found in other simulators is the idea
of maintaining emissions statistics. If the full SimCity code is
available to everyone, the traffic emissions could then be used as
input to tie back into metropolitan flora and fauna health
information. City traffic increases, CO and CO_2 goes up, falcons and
owl birth levels drop, rodent population increases, etc. Its a bit
far fetched, but if the entire system is open source, urban biologists
have a fighting chance.



>> "Don" == Don Hopkins <xardox@mindspring.com> writes:

> Hi, Marc!

> I read about your traffic simulation in your message on
> sourceforge : Project : TheCity : Forums. That sounds quite
> interesting, and is a worthwhile idea. How is it going? I'd like
> to see your traffic site when you publish it.

> I'm working on a educational simulation project with Maxis and
> Columbia University to develop an open source simulation
> programming platform, that would enable projects like yours to be
> easily developed and integrated with other simulators. One aspect
> of the project is taking the original SimCity Classic source code
> (which I have permission from Maxis to use for educational
> projects like this), converting it into a Python module, and
> opening it up with an API, accessors to peek and poke the state,
> and hooks to instrument and override different parts of the
> simulator. I don't have the rights to give out the SimCity source
> code, but I can document and expose its internal state and API
> through Python, and add hooks to make it more useful for
> educational purposes.

> One of the goals is to expose the internal workings of the
> existing traffic simulator (which is quite simplistic), and make
> it possible to plug in new traffic simulators, like your own. Of
> course any traffic simulator that hooked into SimCity would still
> have to deal with its data representations, which are
> fundamentally tile based. But I'm exposing all that through
> Python, so you can overlay and augment it with your own data
> structures and Python objects.

> I'm designing the APIs to hook into different parts of the
> simulator, so I'd appreciate your advice and feedback. How does
> your traffic simulator work? How does it model and interact with
> the environment, and what kinds of inputs and outputs does it
> need? How do you think it could interface with SimCity? What kind
> of an API would you need?

Trafix, from trafix.sourceforge.net, follows basic, simplistic road
traffic algorithms. The algorithms and their history are described
in:

http://etda.libraries.psu.edu/theses/available/etd-0311101-115158/

Pull down the thesis.pdf, which is freely available and see the
traffic theory section. It has a description of the derivation and
history of traffic motion models. The simulation I developed does not
model emissions, it only computes acceleration, velocity, position,
and routes the vehicles through the traffic network. I am not sure
how it would best interact with SimCity. My suggestion at the moment
is to review the Traffic Theory section of my dissertation from the
URL above, and feel free to post me any questions or comments you have
on the work, and how it could be applied to SimCity. There is a large
bibliography which describes the original traffic model sources. If
you are working in Python, it might be easiest to just code it up in
Python.

The real issues in traffic simulation are to develop traffic models,
test them, see how they compare to real traffic patterns and then
adjust and correct the models. Some groups are trying to do this
with automated data collection sensors, but I am not sure if anyone is
mixing an open source project with traffic algorithm model
verification. For more information on my work, see the recent paper::

- An Architecture for a Non-Deterministic Distributed Simulator,
IEEE Transactions on Vehicular Technology, vol. 51, no. 3, May
2002, pp. 453-471.

I think, when you consider that the work started in 1995, and was
submitted for publication before September 11, 2001, you might find it
funny. The original work was designed as a possible traffic
management tool to assist during terrorist incidents in large
metropolitan traffic networks. Thanks for the interest,

Sincerely,

Marc Bumble