Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Car routing
Hi Michael,

I believe that you answered my question unintentionally.

In summary the cars will always looking for the fastast way, correct? So, everything that affect this will change the car decision:
Traffic Jam
Lane Max Speed
Edge Length

There're anything else that I could change via TraCi to affect  this situation?

For my pourposes, the car sensitiveness is helpful because I don't need to worry with them, just change the network status in order to increase the number of arrived cars. The Machine Learn should learn the best situation by itself, considering all variables. However, in order the accelerate the learning process, I can use some heuristics to help the ML choose better actions faster.

Thank you again.


2018-01-16 4:15 GMT-02:00 Michael Behrisch <oss@xxxxxxxxxxx>:
Hi Pedro,
yes the cars use Dijkstra. The grid may be not the best network to test
this because all the edges have the same length and it is thus very
sensitive already to small changes.

Best regards,
Michael

Am 15.01.2018 um 15:52 schrieb Pedro Matuck:
> Hi Michael, thank you for the return. I'll try to be more clear and
> specific.
>
> I'm working in a Machine Learning that should help the traffic flow. In
> summary, I configured a flow to start at some point of network and the
> cars should arrive at the destination. In the middle of that, I
> calculate the number of arrived cars and take an action after some fixed
> number of steps.
>
> What I'm trying to do, is (as you said before) influence the travel time
> changing the network aspects. For instance: At some point of simulation,
> if some lane is with high occupancy, I decrease it 'MaxSpeed' in order
> to make other lanes more interesting for the new inserted cars.
>
> Correct me if I missunderstood, but cars use dijkstra to calculate the
> fastest way, right? I just need to understand the conditions that affect
> the weight of lanes to train my ML in a simple scenario: it involves
> just single lanes, one type of cars and a grid network generated by
> 'netgenerate' (3x3 or 4x4 nodes).
>
> Thank you again.
>
> 2018-01-14 17:44 GMT-02:00 Michael Behrisch <oss@xxxxxxxxxxx
> <mailto:oss@xxxxxxxxxxx>>:
>
>     Hi,
>     the short answer is it takes the fastest route based on the current
>     travel times in the network (this describes the edge part, the lane part
>     is handled by the lane changing algorithm). There are multiple ways to
>     influence which travel times it assumes when calculating the fastest
>     path but maybe you can ask a more specific question here.
>
>     Best regards,
>     Michael
>
>     Am 13.01.2018 um 03:02 schrieb Pedro Matuck:
>     > Hello,
>     >
>     > Please, someone could indicate me the documentation where explain how
>     > cars make a decision when we just declare a flow?
>     >
>     > For example:
>     > I have the following declaration on my *.rou.xml file
>     >
>     > <routes xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance
>     <http://www.w3.org/2001/XMLSchema-instance>"
>     > xsi:noNamespaceSchemaLocation="http://sumo.dlr.de/xsd/routes_file.xsd
>     <http://sumo.dlr.de/xsd/routes_file.xsd>">
>     > <vType id="normal car" vClass="passenger" maxSpeed="40"
>     > speedFactor="0.9" speedDev="0.2" sigma="0.5" color="1,0,0"/>
>     > *<flow id="normal" type="normal car" begin="0" probability="0.2"
>     > from="0/0to1/0" to="1/2to2/2"/>*
>     > </routes>
>     >
>     > After a car is inserted on network, how it makes a decision for which
>     > edge/lane it should take in order to get the destination?
>     >
>     > I read about weights, routing and randomness but nothing were clear enough.
>     >
>     > Thanks.
>     >
>     > --
>     > */Pedro J. Matuck/*
>     > Github <https://github.com/pjmatuck> - Linkedin
>     > <https://www.linkedin.com/in/pedro-matuck-79324323
>     <https://www.linkedin.com/in/pedro-matuck-79324323>> - Youtube
>     > <http://www.youtube.com/c/AbreChaves
>     <http://www.youtube.com/c/AbreChaves>>
>     >
>     > "The hardest battle lies within."
>     >
>     >
>     > _______________________________________________
>     > sumo-user mailing list
>     > sumo-user@xxxxxxxxxxx <mailto:sumo-user@xxxxxxxxxxx>
>     > To change your delivery options, retrieve your password, or
>     unsubscribe from this list, visit
>     > https://dev.eclipse.org/mailman/listinfo/sumo-user
>     <https://dev.eclipse.org/mailman/listinfo/sumo-user>
>     >
>
>
>
>
>
> --
> */Pedro J. Matuck/*
> Github <https://github.com/pjmatuck> - Linkedin
> <https://www.linkedin.com/in/pedro-matuck-79324323> - Youtube
> <http://www.youtube.com/c/AbreChaves>
>
> "The hardest battle lies within."





--
Pedro J. Matuck

"The hardest battle lies within."

Back to the top