Thanks for the example scenario.
The problem is now being tracked here:
https://github.com/eclipse/sumo/issues/4385

2018-08-01 16:19 GMT+02:00 Albiston, Gregory 2012 (PGR) <
[email protected]>:

> Hello Jakob,
>
>
> Thanks for your response.
>
> I've regenerated the network and adjusted pedestrian routing based on your
> advice.
>
> This seems to have fixed most of the negative time losses.
>
>
> However, there seems to still be an issue with the output routeLengths.
>
> Attached is a simple example that includes a pedestrian walking back and
> forth between two points over a crossing, i.e. crossing from one side of a
> road to another.
>
>
> Watching the simulation shows the path taken is the shortest, as expected,
> for both legs.
>
> Yet the output has different routeLengths of 33.20 (shortest path) and
> 173.20 (longer path along the edges).
>
> The duration is identical, as would be expected, resulting in a small and
> a negative time loss respectively.
>
>
> It would seem that the output generation isn't exactly tracking the nodes
> in the route taken during simulation.
>
>
> Thanks,
>
>
> Greg
>
>
> ------------------------------
> *From:* [email protected] <[email protected]> on
> behalf of [email protected] <[email protected]>
> *Sent:* 27 July 2018 13:09
> *To:* [email protected]
> *Subject:* sumo-user Digest, Vol 13, Issue 112
>
> Send sumo-user mailing list submissions to
>         [email protected]
>
> To subscribe or unsubscribe via the World Wide Web, visit
>         https://dev.eclipse.org/mailman/listinfo/sumo-user
> or, via email, send a message with subject or body 'help' to
>         [email protected]
>
> You can reach the person managing the list at
>         [email protected]
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of sumo-user digest..."
>
>
> Today's Topics:
>
>    1. Re: Efficient way to count #vehicles within a distance    of
>       junction (Shuyi Yin)
>    2. Re: Efficient way to count #vehicles within a distance    of
>       junction (Jakob Erdmann)
>    3. Re: Fast Pedestrian Walks/Negative Time Loss (Jakob Erdmann)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Thu, 26 Jul 2018 11:17:02 -0700
> From: Shuyi Yin <[email protected]>
> To: [email protected]
> Subject: Re: [sumo-user] Efficient way to count #vehicles within a
>         distance        of junction
> Message-ID:
>         <CAMYW88ykdm3N+MCatyAAfVrY_A=c94P2ktAULP97WNviKafvGA@mail.
> gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> Hi Harald,
>
> Thank you for letting me know the other approaches: they may be worthwhile
> to explore in the future. And equally thankful for pointing me to that
> discussion.
>
> Sincerely,
> Shuyi
>
> On Thu, Jul 26, 2018 at 12:15 AM Harald Schaefer <
> [email protected]>
> wrote:
>
> > Hi Shuyi,
> >
> > there are different approaches to control traffic lights depending on the
> > traffic.
> > One is to look for gaps in the incoming traffic flow (see
> > http://sumo.dlr.de/wiki/Simulation/Traffic_Lights#
> Improving_Generated_programs_with_knowledge_about_traffic_demand
> > and
> > http://sumo.dlr.de/wiki/Simulation/Traffic_Lights#
> Actuated_Traffic_Lights)
> > For your approach see also the discussion recently here about "Locating
> > cars around certain coordinates"
> >
> > Best regards, Harald
> >
> > Am 25.07.2018 um 23:43 schrieb Shuyi Yin:
> >
> > Hi SUMO community,
> >
> > I'm using TraCI to control traffic lights and would like the traffic
> light
> > to know the number of cars approaching it. Also, I'm only interested in
> > counting vehicles that are close enough, i.e. within 20 meters.
> >
> > My current implementation is to get the ID of vehicles on a lane and
> > calculate the distance to the junction. Code is shown as follows. But is
> > there a more efficient way fo doing so?
> >
> > Thanks,
> > Shuyi
> >
> > def numberv(lane):
> >     nb = 0
> >     for k in traci.lane.getLastStepVehicleIDs(lane):
> >         if traci.vehicle.getLanePosition(k) < X-100:
> >              nb += 1
> >     return nb
> >
> >
> >
> >
> > _______________________________________________
> > sumo-user mailing [email protected]
> > To change your delivery options, retrieve your password, or unsubscribe
> from this list, visithttps://dev.eclipse.org/mailman/listinfo/sumo-user
> >
> >
> > _______________________________________________
> > sumo-user mailing list
> > [email protected]
> > To change your delivery options, retrieve your password, or unsubscribe
> > from this list, visit
> > https://dev.eclipse.org/mailman/listinfo/sumo-user
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <https://dev.eclipse.org/mailman/private/sumo-user/
> attachments/20180726/bd59cedc/attachment.html>
>
> ------------------------------
>
> Message: 2
> Date: Fri, 27 Jul 2018 08:52:28 +0200
> From: Jakob Erdmann <[email protected]>
> To: Sumo project User discussions <[email protected]>
> Subject: Re: [sumo-user] Efficient way to count #vehicles within a
>         distance        of junction
> Message-ID:
>         <CAMbUcbymYxk52_obxWZf8xanywLrOe_qz2bAhROuCevTDbh3qw@mail.
> gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> I was thinking of placing a 20m long detector on each approaching lane to
> cover the area of interest and count the cars therein (and also measure
> speed and flow as needed).
>
> 2018-07-26 20:14 GMT+02:00 Shuyi Yin <[email protected]>:
>
> > Hi Jakob,
> >
> > Thank you for replying. In terms of E2-Detector, are you suggesting that
> I
> > put a 20m-long E2-detector on the lane to measure the number of vehicles,
> > or several E2-detectors connected to cover the 20m distance?
> >
> > Or are you saying putting one normal E2-detector to monitor the flow, but
> > not the number of vehicles in a 20m-long area directly?
> >
> > Thanks,
> > Shuyi
> >
> > On Thu, Jul 26, 2018 at 12:50 AM Jakob Erdmann <[email protected]>
> > wrote:
> >
> >> In my own control scripts I often use explicit list of lanes but you
> >> could also use
> >> E2-Detectors generated in advance http://sumo.dlr.de/wiki/Tools/
> >> Output#generateTLSE2Detectors.py
> >>
> >>
> >> 2018-07-26 9:14 GMT+02:00 Harald Schaefer <[email protected]>:
> >>
> >>> Hi Shuyi,
> >>>
> >>> there are different approaches to control traffic lights depending on
> >>> the traffic.
> >>> One is to look for gaps in the incoming traffic flow (see
> >>> http://sumo.dlr.de/wiki/Simulation/Traffic_Lights#
> >>> Improving_Generated_programs_with_knowledge_about_traffic_demand and
> >>> http://sumo.dlr.de/wiki/Simulation/Traffic_Lights#
> >>> Actuated_Traffic_Lights)
> >>> For your approach see also the discussion recently here about "Locating
> >>> cars around certain coordinates"
> >>>
> >>> Best regards, Harald
> >>>
> >>>
> >>> Am 25.07.2018 um 23:43 schrieb Shuyi Yin:
> >>>
> >>> Hi SUMO community,
> >>>
> >>> I'm using TraCI to control traffic lights and would like the traffic
> >>> light to know the number of cars approaching it. Also, I'm only
> interested
> >>> in counting vehicles that are close enough, i.e. within 20 meters.
> >>>
> >>> My current implementation is to get the ID of vehicles on a lane and
> >>> calculate the distance to the junction. Code is shown as follows. But
> is
> >>> there a more efficient way fo doing so?
> >>>
> >>> Thanks,
> >>> Shuyi
> >>>
> >>> def numberv(lane):
> >>>     nb = 0
> >>>     for k in traci.lane.getLastStepVehicleIDs(lane):
> >>>         if traci.vehicle.getLanePosition(k) < X-100:
> >>>              nb += 1
> >>>     return nb
> >>>
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> sumo-user mailing [email protected]
> >>> To change your delivery options, retrieve your password, or
> unsubscribe from this list, visithttps://dev.eclipse.org/
> mailman/listinfo/sumo-user
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> sumo-user mailing list
> >>> [email protected]
> >>> To change your delivery options, retrieve your password, or unsubscribe
> >>> from this list, visit
> >>> https://dev.eclipse.org/mailman/listinfo/sumo-user
> >>>
> >>>
> >> _______________________________________________
> >> sumo-user mailing list
> >> [email protected]
> >> To change your delivery options, retrieve your password, or unsubscribe
> >> from this list, visit
> >> https://dev.eclipse.org/mailman/listinfo/sumo-user
> >>
> >
> > _______________________________________________
> > sumo-user mailing list
> > [email protected]
> > To change your delivery options, retrieve your password, or unsubscribe
> > from this list, visit
> > https://dev.eclipse.org/mailman/listinfo/sumo-user
> >
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <https://dev.eclipse.org/mailman/private/sumo-user/
> attachments/20180727/e484b2df/attachment.html>
>
> ------------------------------
>
> Message: 3
> Date: Fri, 27 Jul 2018 14:09:22 +0200
> From: Jakob Erdmann <[email protected]>
> To: Sumo project User discussions <[email protected]>
> Subject: Re: [sumo-user] Fast Pedestrian Walks/Negative Time Loss
> Message-ID:
>         <CAMbUcbwy=nh6OXearZDBc-4Z8ro8E2E+RbofEH9ZHXvS5sFuvg@
> mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
>
> Hello,
> the problematic behavior is caused by an invalid pedestrian route.
> The edge sequence 633 -629 leads to junction 476 which is not connected
> with next route edge -813
> This causes the person to walk contrary to the expected direction and thus
> reach the arrival position earlier than expected.
> In the latest development version of SUMO additional warnings have been
> implemented to make the faulty behavior more obvious.
> A possible fix would be to either use the sequence 633 -629 -629 -813 or
> 633 -813
>
> The network itself is problematic insofer that it deviates from the two
> possible styles for pedestrian network modelling:
> 1) with crossings and walkingAreas. This makes it explicit where
> pedestrians may cross the road and is the recommended style for detailed
> pedestrian simulation (
> http://sumo.dlr.de/wiki/Simulation/Pedestrians#Building_a_network_for_
> pedestrian_simulation
> )
> 2) without crossing and walkingAreas but with sidewalks that have no
> connections. In this case, pedestrians will assume a fully connected
> topology at each intersection but their paths are not modelled (they "jump"
> across). This type of network is generated automatically by netconvert when
> not generating crossings or walkingareas. Note, that "jumping" potentially
> causes negative timeLoss in the current version but this will only be
> noticable if the dwadling option is disabled.
>
> In your case the network contains sidewalks that are connected but this
> forces awkward detours because there are no connections that allow changing
> the walking direction.
>
> regards,
> Jakob
>
>
>
>
>
>
> 2018-07-25 15:15 GMT+02:00 Albiston, Gregory 2012 (PGR) <
> [email protected]>:
>
> > Hello,
> >
> >
> > I have a scenario where there are numerous pedestrians undertaking
> walking
> > stages.
> >
> > Some of these stages are resulting in negative time losses meaning the
> > pedestrian is arriving faster than the maximum speed.
> >
> > All the vehicle stages have positive time losses.
> >
> >
> > Attached is an example scenario for one pedestrian which shows the issue.
> > The first and final stages are completed with a time loss of -79.75 and
> > -188.83. This equates to speeds of 1.84 and 2.41 when the max speed is
> set
> > to 1.79.
> >
> >
> > Is there a modelling reason for the negative time losses?
> >
> > Apologies if I've missed something.
> >
> >
> > Thanks,
> >
> >
> > Greg
> > DISCLAIMER: This email is intended solely for the addressee. It may
> > contain private and confidential information. If you are not the intended
> > addressee, please take no action based on it nor show a copy to anyone.
> In
> > this case, please reply to this email to highlight the error. Opinions
> and
> > information in this email that do not relate to the official business of
> > Nottingham Trent University shall be understood as neither given nor
> > endorsed by the University. Nottingham Trent University has taken steps
> to
> > ensure that this email and any attachments are virus-free, but we do
> advise
> > that the recipient should check that the email and its attachments are
> > actually virus free. This is in keeping with good computing practice.
> >
> > _______________________________________________
> > sumo-user mailing list
> > [email protected]
> > To change your delivery options, retrieve your password, or unsubscribe
> > from this list, visit
> > https://dev.eclipse.org/mailman/listinfo/sumo-user
> >
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <https://dev.eclipse.org/mailman/private/sumo-user/
> attachments/20180727/e8c2356a/attachment.html>
>
> ------------------------------
>
> _______________________________________________
> sumo-user mailing list
> [email protected]
> To change your delivery options, retrieve your password, or unsubscribe
> from this list, visit
> https://dev.eclipse.org/mailman/listinfo/sumo-user
>
>
> End of sumo-user Digest, Vol 13, Issue 112
> ******************************************
> DISCLAIMER: This email is intended solely for the addressee. It may
> contain private and confidential information. If you are not the intended
> addressee, please take no action based on it nor show a copy to anyone. In
> this case, please reply to this email to highlight the error. Opinions and
> information in this email that do not relate to the official business of
> Nottingham Trent University shall be understood as neither given nor
> endorsed by the University. Nottingham Trent University has taken steps to
> ensure that this email and any attachments are virus-free, but we do advise
> that the recipient should check that the email and its attachments are
> actually virus free. This is in keeping with good computing practice.
>
> _______________________________________________
> sumo-user mailing list
> [email protected]
> To change your delivery options, retrieve your password, or unsubscribe
> from this list, visit
> https://dev.eclipse.org/mailman/listinfo/sumo-user
>
>
_______________________________________________
sumo-user mailing list
[email protected]
To change your delivery options, retrieve your password, or unsubscribe from 
this list, visit
https://dev.eclipse.org/mailman/listinfo/sumo-user

Reply via email to