Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Realistic Simulation with traffic data and calibration
Hello,
- there should be no difference in vehicle behavior between running a scenario with sumo or sumo-gui
- there is a known issue when trying to re-run a calibrator scenario without the calibrators (https://github.com/eclipse/sumo/issues/5826)
- if you have turn-count data, I would recommend using the new routeSampler tool. It works similar to calibrators but calibrates the routes in an offline manner so you can then rerun it without calibrators. Make sure to use the latest development version because it includes additional options for optimizing the calibration with an ILP solver (https://en.wikipedia.org/wiki/Integer_programming). See https://sumo.dlr.de/docs/Tools/Turns.html#routesamplerpy

regards,
Jakob

Am Mo., 2. März 2020 um 11:27 Uhr schrieb Tetris <schmelter@xxxxxxxxxxxxxxxxxxx>:
Hello,

my goal is a realistic simulation of a part of a city (downtown, no highway)
where my vehicle will drive a certain route so that i can test how fast it
gets from A to B. I have traffic data of all the junctions that the vehicle
will cross. The traffic data consists of vehicle counts (every hour is
summed up) from every road X to every reachable road Y of the junction.

I got my network by using the osmWebWizard and deleting areas that are
outside of my interest.
I used randomtrips.py to get random routes in my network, then calibration
with detectors.

What I have now is an output file with route distributions and a file with
detector output.
The detector output shows me that nVehContrib and aspiredFlow are not the
same. aspiredFlow are the same values as vehsPerHour in my calibrator.xml.
What I did then is tweaking the p parameter of randomtrips.py to get
nVehContrib to the correct count of vehicles (vehsPerHour).

I also get many warnings that certain vehicles couldn't be removed. So the
calibrators couldn't work properly.
Warning: Calibrator 'xyz' could not remove vehicle 'abc'.
But my nVehContrib is way too low!

Another warning is:
Teleporting vehicle 'xyz'; waited too long (jam).


Is this the way I should go on or is DFROUTER a better way? I thought for
meshed cities calibration should be used as it is stated in the wiki.


Another question I have is how can I run the simulation in sumo gui after I
calibrated it. If I run the simulation in sumo GUI while calibration I get
other results than by using command line. That's why I thought I would use
the route distribution after calibration to run the sumo GUI. But I don't
know how. There are no vehicles in my route distribution file.
 It's only like this:
    <routeDistribution id="xyz">
        <route id="abc" edges=".. ... ..." probability="1.00"/>
    </routeDistribution>


This is part of my calibration.xml. There is a t0 in every flow:
    <vType id="t0" accel="0.8" decel="4.5" sigma="0.5" length="5"
maxSpeed="70" speedDev="0.1"/>

      <flow id=".." begin="0"    end="3600" route="..." vehsPerHour="28"
speed="13.8" type="t0" departPos="free" departSpeed="max"/>
    </calibrator>


How can I match every route distribution with a vehicle type by using
command line?

I know your time is valuable and I appreciate your attention.



--
Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-user

Back to the top