It is only suspicious in the sense of being unexpected. If your algorithm
was supposed to generate routes that are more natural, this would point to
a problem with the algorithm. In the very least, routes with frequent
turn-arounds are more likely to generate traffic jams due to the low
right-of-way priority when performing a turn-around and the potential to
block the following traffic. If many vehicle have these looping routes this
can be likened to a roundabout where the inflow edges have right-of-way. In
heavy traffic this is prone to dead-lock.
If vehicles have no clear objective, they could be moved outside the road
network by giving them a <stop> with parking=true.
If they should remain on the road it could be preferable to give them a
random destination some else in the network.
regards,
Jakob
2017-02-22 12:08 GMT+01:00 Green Mind <[email protected]>:
> In my simulation I have vehicles that sometimes need to go into
> 'wandering' state, so I generate random routes for them in runtime using a
> loop of *random.choice(e.getOutgoing())* (where 'e' is the previous edge
> in the route).
>
> It does lead sometimes to vehicles driving in circles over opposing
> directions of the same street but this is fine for my purposes.
>
> Why is it then 'suspicious' and is there a better way you suggest I use?
>
> On Wed, Feb 22, 2017 at 9:13 AM, Jakob Erdmann <[email protected]
> > wrote:
>
>> Your example runs fine with 0.29.0 and also with the latest development
>> version. The routes look somewhat suspicious though, because they include
>> many tight loops.
>>
>> regards,
>> Jakob
>>
>> 2017-02-21 18:00 GMT+01:00 Green Mind <[email protected]>:
>>
>>> Sorry, I know see that this issue is marked as fixed for version 0.29.0.
>>> I will check the new version out first and update you with the results.
>>>
>>> On Tue, Feb 21, 2017 at 6:57 PM, Green Mind <[email protected]>
>>> wrote:
>>>
>>>> Hi,
>>>>
>>>> I attach another set of input files which cause deadlocks, even after I
>>>> set roundabouts.guess in netedit. I also tried to rebuild the network using
>>>> recommended NETCONVERT options, and the issue still happened.
>>>>
>>>> Will you please take a look? This is becoming a real blocker in our
>>>> project, as we don't have a solution for predicting and working around
>>>> these deadlocks :(
>>>>
>>>> Thanks.
>>>>
>>>> On Tue, Jan 31, 2017 at 9:41 AM, Jakob Erdmann <
>>>> [email protected]> wrote:
>>>>
>>>>> In the previous example I noticed that your vehicle does lots of
>>>>> little turnaround loops (driving over the same edge multiple times). I
>>>>> suspect that this is part of the reason for the deadlocks you see.
>>>>> However, it's not obvious why the vehicles in your screenshot would be
>>>>> stopped. Please send me the input files again ^^
>>>>>
>>>>> regards,
>>>>> Jakob
>>>>>
>>>>> 2017-01-30 22:19 GMT+01:00 Green Mind <[email protected]>:
>>>>>
>>>>>> Jakob,
>>>>>>
>>>>>> I tried your suggestion and rebuilt the network with all the recommended
>>>>>> NETCONVERT options
>>>>>> <http://sumo.dlr.de/wiki/Networks/Import/OpenStreetMap#Recommended_NETCONVERT_Options>.
>>>>>> While there might be a slight improvement, I still get often "deadlocks"
>>>>>> (see pictures below - all the 6 vehicles in the pictures are stuck and
>>>>>> don't move, although they have a long route to follow). Any ideas?
>>>>>>
>>>>>> [image: Inline image 2]
>>>>>>
>>>>>> [image: Inline image 1]
>>>>>>
>>>>>> On Mon, Jan 30, 2017 at 10:55 PM, Green Mind <[email protected]>
>>>>>> wrote:
>>>>>>
>>>>>>> Thanks Jakob.
>>>>>>>
>>>>>>> Is there a way to workaround this issue without rebuilding the
>>>>>>> network?
>>>>>>> I couldn't find how do it with netedit.
>>>>>>>
>>>>>>> On Mon, Jan 30, 2017 at 1:37 PM, Jakob Erdmann <
>>>>>>> [email protected]> wrote:
>>>>>>>
>>>>>>>> Hello,
>>>>>>>> thank you for pointing out this bug. It is now being tracked here:
>>>>>>>> http://sumo.dlr.de/trac.wsgi/ticket/2834
>>>>>>>> As a simple workaround, rebuild your network using the option
>>>>>>>> --roundabouts.guess (can also be set in Netedit).
>>>>>>>>
>>>>>>>> regards,
>>>>>>>> Jakob
>>>>>>>>
>>>>>>>> 2017-01-30 10:37 GMT+01:00 Green Mind <[email protected]>:
>>>>>>>>
>>>>>>>>> Hi all,
>>>>>>>>>
>>>>>>>>> I have a scenario with a simple map imported from OSM, where a
>>>>>>>>> vehicle gets
>>>>>>>>> stuck inside an intersection while trying to turn right. The route
>>>>>>>>> is
>>>>>>>>> generated randomly using a script. I use Sumo version 0.28.0
>>>>>>>>>
>>>>>>>>> This obviously screws up my entire simulation, because all the
>>>>>>>>> vehicles who
>>>>>>>>> arrive behind it are also unable to turn anywhere (they're waiting
>>>>>>>>> for the
>>>>>>>>> stuck vehicle in front of them to turn).
>>>>>>>>>
>>>>>>>>> I attach a screenshot, as well as the simulation files. Would be
>>>>>>>>> glad for
>>>>>>>>> any clues on how to solve this.
>>>>>>>>>
>>>>>>>>> TIA!
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> ------------------------------------------------------------
>>>>>>>>> ------------------
>>>>>>>>> Check out the vibrant tech community on one of the world's most
>>>>>>>>> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
>>>>>>>>> _______________________________________________
>>>>>>>>> sumo-user mailing list
>>>>>>>>> [email protected]
>>>>>>>>> https://lists.sourceforge.net/lists/listinfo/sumo-user
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
sumo-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sumo-user
