[BonnMotion] OSRM Installation
Matthias Schwamborn
schwamborn at informatik.uos.de
Wed Oct 11 17:04:37 CEST 2017
Fátima,
the movement heavily depends on your map extract, the types of OSM
highways (cf. [1]) within, and the resulting movement graph. From your
problem description, I can only guess that a lot of routes require the
nodes to leave the simulation area since the globally shortest route is
not contained within it. You might want to try to use the extract
defined by your bounding box as input for both BonnMotion and OSRM. This
should result in movement within the simulation area only.
Note that this is all assuming the tested older version of OSRM. There
might also have been changes in the newer OSRM version which cause
unwanted behavior.
Best regards,
Matthias
[1] http://wiki.openstreetmap.org/wiki/Key:highway
On 19.09.17 16:51, Fátima Castro Jul wrote:
> Thanks for the reference to the paper. It has provided me with a better
> understanding of the tool but still not complete. I am still not able to
> generate the scenario I desire. I am not sure if it is related to the
> limitations of scenario generation in Bonnmotion or to the fact I have
> not correctly understood the results I should expect. I am mentioning
> the tool limitations because in every paper I have found where
> Bonnmotion is employed, areas are bigger than mine and fewer nodes are
> generated. Is it unreasonable to use the tool to generate 100 or 500
> nodes in an area of less than 2 km^2 ? As far as I understand, such
> pedestrian density would make sense in a real scenario. When I try
> generate it, I am unable to obtain a scenario where nodes stay in the
> simulation area and their movements are correct. The problems I
> experience are the following:
>
> * For C=0, every node is outside the simulation area (default -i
> 3600). Using -i 0 nodes do have their starting position in the area
> but are soon moving outside (and far of) the bounding box. I
> understand this behavior is due to the small size of the area and,
> therefore, is reasonable.
> * For C=1, nodes are in the simulation area and move as expected but a
> significant number of them concentrate in one of the limits of the
> bounding box. This is not desirable. I guess it is due to their
> routes being cut but I do not understand why they only concentrate
> on one of the limits of the area. Moreover, I wonder if this is
> expected to happen always with C = 1 or it is a consequence of the
> high amount of nodes generated.
> * For C= 2, nodes are in the simulation area and do not concentrate on
> any of the edges but their routes are too short (start and
> destination points are really close) and therefore they are barely
> moving. This is also undesirable. Is is due to the small size of the
> area?
>
> Best regards,
>
> Fátima
>
>
>
> On 15/09/17 11:42, Matthias Schwamborn wrote:
>> Please read up on the meaning of the -C parameter in [1, Section III-B].
>>
>> Best regards,
>> Matthias
>>
>> [1] http://ieeexplore.ieee.org/abstract/document/7330189/
>>
>> On 14.09.17 19:07, Fátima Castro Jul wrote:
>>> I have realised I was using the parameter -C 0 (CLIP_RESIZE_BB) as it
>>> was set by default. As a result, the bounding box that limits the
>>> simulation area was expanded. I guess that because my area is small
>>> almost every point was outside it.
>>>
>>> I have run Bonnmotion changing the parameter value. With -C 1
>>> (CLIP_USE_NEW_LENGTH) nothing happens outside the box but many nodes are
>>> crowded together on one of the limits of the simulation area. With -C 2
>>> (CLIP_USE_OLD_LENGTH) the distributions is better, there are some groups
>>> of nodes near the edges but it is not that crowded. However, in this
>>> scenario, nodes are not moving at all. Is this the normal behavior of
>>> this parameter?
>>>
>>> Best regards,
>>>
>>> Fátima
>>>
>>>
>>> On 14/09/17 16:53, Fátima Castro Jul wrote:
>>>> Hi,
>>>>
>>>> Thank you for your answer.
>>>>
>>>> I have tried to modify OSRoutingMachine.java//according to the new
>>>> OSRM API to work with ORM 5.11.0. My modifications are limited to the
>>>> API calls and responses processing. Now, I can run Bonnmotion.
>>>> However, the coordinates in the .movements.gz file are not within the
>>>> limits of the x,y values in the//.params file. The coordinates in the
>>>> file .movements.geo.gz are not always within the bounding box either.
>>>>
>>>> May the modifications to the OSRoutingMachine.java file have affected
>>>> coordinates choosing? Am I doing something wrong when running Bonnmotion?
>>>>
>>>> For instance:
>>>>
>>>> ./bm -f scenario RandomStreet -n 1000 -d 3600 -s 0.9 1.5 -p 30 -B
>>>> -8.7296 42.2319 -8.7139 42.2421 -R 1234 -o /path/map.osm.pbf -u
>>>> 0.0.0.0:5001
>>>>
>>>>
>>>> In scenario.params:
>>>>
>>>> model=RandomStreet
>>>> ignore=3600.0
>>>> randomSeed=1234
>>>> x=1747.716005454422
>>>> y=1533.635604236275
>>>> duration=3600.0
>>>> nn=1000
>>>> circular=false
>>>> J=2D
>>>> boundingBox=-8.7296 42.2319 -8.7139 42.2421
>>>> clippingMethod=0
>>>> rsMetric=Pedestrian
>>>> osmFile=/path/map.osm.pbf
>>>> rsURL=0.0.0.0:5001
>>>> distFile=null
>>>> maxPause=30.0
>>>> speed=0.9 1.5
>>>>
>>>>
>>>> In scenario.movements.geo.gz (first lines):
>>>>
>>>> 0.0 [-8.715823405291129,42.203754445654724], 3600.0
>>>> [-8.706116197524537,42.1687900700981],
>>>> 0.0 [-8.718972344559301,42.203802813972466], 3600.0
>>>> [-8.71099292724169,42.17492810899706],
>>>> 0.0 [-8.71079316577188,42.19073929147142], 3600.0
>>>> [-8.69885787606997,42.14771951347391],
>>>> 0.0 [-8.706779739246365,42.20395795031393], 3600.0
>>>> [-8.698081948572195,42.17259861887113],
>>>>
>>>> In scenario.movements.gz (first lines):
>>>>
>>>> 0.0 1187.5624645399644 -3327.2085444231298 3600.0 1920.5450429319296
>>>> -7672.257377061738
>>>> 0.0 896.5370442566041 -3471.3099859895324 3600.0 1498.2509550961317
>>>> -7057.227789234152
>>>> 0.0 1665.0420254702763 -5067.26831240366 3600.0 2566.700416711787
>>>> -10413.153339571183
>>>> 0.0 2228.608934641187 -3385.276691569191 3600.0 2885.3050883430005
>>>> -7282.212748053645
>>>>
>>>> Best regards,
>>>>
>>>> Fátima
>>>>
>>>>
>>>>
>>>> On 06/09/17 14:10, Matthias Schwamborn wrote:
>>>>> Hi Fátima,
>>>>>
>>>>>> I would appreciate if you could confirm the folder you have hinted me to look into, considering the error I encounter is in a different one.
>>>>>> ~/bonnmotion-3.0.1/bin$ ./bm -f Vigo-escenario MSLAW -n 5
>>>>> there are several mandatory parameters for map-based scenarios missing
>>>>> here, e.g., -B, -o, -u. Please refer to the documentation [1, Section
>>>>> 6.7.1] for details.
>>>>>
>>>>>
>>>>> Best, Matthias
>>>>>
>>>>> [1] http://sys.cs.uos.de/bonnmotion/doc/README.pdf
>>>>>
>>>>> On 06.09.17 13:15, Fátima Castro Jul wrote:
>>>>>> Hi,
>>>>>>
>>>>>> Thank you for your answer.
>>>>>>
>>>>>> I would appreciate if you could confirm the folder you have hinted me to
>>>>>> look into, considering the error I encounter is in a different one.
>>>>>>
>>>>>> Best regards,
>>>>>> Fátima
>>>>>>
>>>>>>
>>>>>> On 05/09/17 18:05, Matthias Schwamborn wrote:
>>>>>>> Hi Fátima,
>>>>>>>
>>>>>>> the short answer is that the current implementation of BonnMotion
>>>>>>> assumes the route request and response formats of OSRM version 4.8.1 and
>>>>>>> this format has been changed OSRM-side since then at least once.
>>>>>>>
>>>>>>> As much as I'd like to update the BM-side of the code, there are several
>>>>>>> reasons why we haven't, yet, and most likely won't in the near future.
>>>>>>> However, if you are eager enough to get BM working with OSRM version
>>>>>>> 5.11, you can dig into and update the code located in
>>>>>>> 'src/edu/bonn/cs/iv/util/maps/OSRoutingMachine.java'.
>>>>>>>
>>>>>>> As for why you encounter that error with version OSRM version 4.8.1, I
>>>>>>> can only assume that this version might be incompatible with some newer
>>>>>>> libraries in Ubuntu 17.04.
>>>>>>>
>>>>>>>
>>>>>>> Best regards,
>>>>>>> Matthias
>>>>>>>
>>>>>>> On 05.09.17 17:34, Fátima Castro Jul wrote:
>>>>>>>> Hi again!
>>>>>>>>
>>>>>>>> I have been able to set up and run osrm-backend-5.11 with some
>>>>>>>> modifications to the instructions in the documentation.
>>>>>>>>
>>>>>>>> * I have replaced cbf-routing-profiles for the newest ones in
>>>>>>>> https://github.com/sosm/cbf-routing-profiles/
>>>>>>>> * I have changed the last line in prepare_pbf.sh to the following:
>>>>>>>> LUA_PATH="$profiledir/lib/?.lua" ./osrm-contract ${osmfile}.osrm
>>>>>>>>
>>>>>>>> However, I have not been able to generate a map-base scenario
>>>>>>>> successfully. Both with RandomStreet and MSLAW I run into the following
>>>>>>>> error:
>>>>>>>>
>>>>>>>> ~/bonnmotion-3.0.1/bin$ ./bm -f Vigo-escenario MSLAW -n 5
>>>>>>>> BonnMotion 3.0.1
>>>>>>>>
>>>>>>>> OS: Linux 4.10.0-33-generic
>>>>>>>> Java: Oracle Corporation 1.8.0_131
>>>>>>>>
>>>>>>>> Starting MSLAW ...
>>>>>>>> DEBUG: transformation (proj4): +proj=merc +lon_0=0 +k=1 +x_0=0
>>>>>>>> +y_0=0 +a=6378137 +b=6378137 +towgs84=0,0,0,0,0,0,0 +units=m
>>>>>>>> +no_defs
>>>>>>>> java.lang.reflect.InvocationTargetException
>>>>>>>> at
>>>>>>>> sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native
>>>>>>>> Method)
>>>>>>>> at
>>>>>>>>
>>>>>>>> sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
>>>>>>>>
>>>>>>>> at
>>>>>>>>
>>>>>>>> sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
>>>>>>>>
>>>>>>>> at
>>>>>>>> java.lang.reflect.Constructor.newInstance(Constructor.java:423)
>>>>>>>> at edu.bonn.cs.iv.bonnmotion.run.BM.go(BM.java:267)
>>>>>>>> at edu.bonn.cs.iv.bonnmotion.run.BM.main(BM.java:296)
>>>>>>>> Caused by: java.lang.NullPointerException
>>>>>>>> at
>>>>>>>>
>>>>>>>> edu.bonn.cs.iv.bonnmotion.MapScenario.preGeneration(MapScenario.java:423)
>>>>>>>>
>>>>>>>> at
>>>>>>>>
>>>>>>>> edu.bonn.cs.iv.bonnmotion.models.MSLAW.preGeneration(MSLAW.java:631)
>>>>>>>> at
>>>>>>>> edu.bonn.cs.iv.bonnmotion.models.MSLAW.generate(MSLAW.java:101)
>>>>>>>> at edu.bonn.cs.iv.bonnmotion.models.MSLAW.go(MSLAW.java:96)
>>>>>>>> at edu.bonn.cs.iv.bonnmotion.models.MSLAW.<init>(MSLAW.java:90)
>>>>>>>> ... 6 more
>>>>>>>>
>>>>>>>> Error:
>>>>>>>> Error in MSLAW
>>>>>>>>
>>>>>>>> Is there a path where osrm is required to be running so it can work with
>>>>>>>> Bonnmotion? Or is there any other requirement so they can work together?
>>>>>>>> Have I compromised Bonnmotion function due to my modifications to the
>>>>>>>> osrm setting up process?
>>>>>>>>
>>>>>>>> Best regards,
>>>>>>>> Fátima
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> BonnMotion mailing list
>>>>>>>> BonnMotion at list.serv.Uni-Osnabrueck.de
>>>>>>>> https://list.serv.uni-osnabrueck.de/mailman/listinfo/bonnmotion
>>>>>>>>
>
--
Matthias Schwamborn
University of Osnabrück Tel.: +49-541-969-7167
Institute of Computer Science Fax: +49-541-969-2799
Wachsbleiche 27 E-mail: schwamborn at informatik.uos.de
D-49090 Osnabrück, Germany http://sys.cs.uos.de/schwamborn/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 203 bytes
Desc: OpenPGP digital signature
URL: <http://list.serv.uni-osnabrueck.de/pipermail/bonnmotion/attachments/20171011/0cf439d1/attachment.sig>
More information about the BonnMotion
mailing list