> Hi Ugo,
>
> For experiments, we generally use the REDUCED_MODEL_SET which has 15
> models. At that order, you should definitely not exceed 7Gb. When
> trying more than 30 models though we ran into similar problems, like
> Peter mentioned.
>
> For the next release we are hoping to provide a remote service running
> on one of our own machines that will have many models already loaded
> in memory, but that is still in the works and probably won't be out
> for another couple of weeks at earliest...
>
> Matei
>
> On Mon, Dec 6, 2010 at 1:45 AM, Ugo Cupcic <ugo@shadowrobot.com
> <mailto:ugo@shadowrobot.com>> wrote:
>
> Hi Peter,
>
> Thanks for your answer, I'll look into that and try to build a
> subset of the models.
>
> By the way, it seems to be taking a lot more than 2Gb (I tried to
> load it and it took more than 7Gb (and then crashed)).
>
> Cheers,
>
> Ugo
>
>
> On 05/12/10 00:26, Peter Brook wrote:
>> Hey Ugo,
>>
>> In my experience with the tabletop_object_detector, the memory
>> usage can be quite bad. The usage depends on the number of models
>> in the model_set, but I have seen the node use over 2GB of memory.
>>
>> You can change the number of models being loaded by creating a
>> new model set with fewer models than the default one. Since you
>> have your own local copy of the database, take a look at the
>> model_set table of household_objects_database.
>>
>> Peter
>>
>> On Thu, Dec 2, 2010 at 8:51 AM, Ugo Cupcic <ugo@shadowrobot.com
>> <mailto:ugo@shadowrobot.com>> wrote:
>>
>> Hi all,
>>
>> I'm trying to use the tabletop_object_detector. It works fine
>> when I'm
>> setting use_database to false.
>>
>> The problem I have is when I'm trying to use the database: I
>> installed a
>> local copy of the household_objects_database. The
>> household_objects_database node seems to be working fine.
>> However, when
>> I start the tabletop_object_detector/tabletop_node, the node
>> crashes
>> while loading the models (it loads lots of models before
>> crashing).
>>
>> Any idea ?
>>
>> I also had an other quick question: how much RAM does the
>> tabletop_node
>> need to load the database? With 4Gb I'm swapping constantly.
>>
>> You can find the backtrace below.
>>
>> Cheers,
>>
>> Ugo
>>
>> [ INFO] [1291308286.662052341]: Loaded database model with
>> id 18737
>> [ INFO] [1291308289.346903337]: Loaded database model with
>> id 18736
>> terminate called after throwing an instance of 'std::bad_alloc'
>> what(): std::bad_alloc
>>
>> Program received signal SIGABRT, Aborted.
>> 0xb7fe2430 in __kernel_vsyscall ()
>> (gdb) backtrace
>> #0 0xb7fe2430 in __kernel_vsyscall ()
>> #1 0xb7238651 in raise () from /lib/tls/i686/cmov/libc.so.6
>> #2 0xb723ba82 in abort () from /lib/tls/i686/cmov/libc.so.6
>> #3 0xb748552f in __gnu_cxx::__verbose_terminate_handler() ()
>> from /usr/lib/libstdc++.so.6
>> #4 0xb7483465 in ?? () from /usr/lib/libstdc++.so.6
>> #5 0xb74834a2 in std::terminate() () from
>> /usr/lib/libstdc++.so.6
>> #6 0xb74835e1 in __cxa_throw () from /usr/lib/libstdc++.so.6
>> #7 0xb7483c5f in operator new(unsigned int) () from
>> /usr/lib/libstdc++.so.6
>> #8 0xb7483d3d in operator new[](unsigned int) () from
>> /usr/lib/libstdc++.so.6
>> #9 0xb7a7f616 in VoxelGrid (this=0x1cd82d0,
>> size_x=0.28061000210046771,
>> size_y=0.27929999691247942, size_z=0.12600000074505804,
>> resolution=0.002,
>> origin_x=-0.14644000318646433, origin_y=-0.14591999891400337,
>> origin_z=-0.0050000000000000001, default_object=...)
>> at
>> /tmp/buildd/ros-cturtle-collision-environment-0.2.2/debian/ros-cturtle-collision-environment/opt/ros/cturtle/stacks/collision_environment/distance_field/include/distance_field/voxel_grid.h:193
>> #10 DistanceField (this=0x1cd82d0, size_x=0.28061000210046771,
>> size_y=0.27929999691247942, size_z=0.12600000074505804,
>> resolution=0.002,
>> origin_x=-0.14644000318646433, origin_y=-0.14591999891400337,
>> origin_z=-0.0050000000000000001, default_object=...)
>> at
>> /tmp/buildd/ros-cturtle-collision-environment-0.2.2/debian/ros-cturtle-collision-environment/opt/ros/cturtle/stacks/collision_environment/distance_field/---Type
>> <return> to continue, or q <return> to quit---
>> include/distance_field/distance_field.h:145
>> #11 0xb7a7c53b in PropagationDistanceField (this=0x1cd82d0,
>> size_x=0.28061000210046771, size_y=0.27929999691247942,
>> size_z=0.12600000074505804, resolution=0.002,
>> origin_x=-0.14644000318646433, origin_y=-0.14591999891400337,
>> origin_z=-0.0050000000000000001,
>> max_distance=0.10000000149011612)
>> at
>> /tmp/buildd/ros-cturtle-collision-environment-0.2.2/debian/ros-cturtle-collision-environment/opt/ros/cturtle/stacks/collision_environment/distance_field/src/propagation_distance_field.cpp:49
>> #12 0x080d42f5 in
>> tabletop_object_detector::DistanceFieldFitter::initializeFromBtVectors
>> (this=0x2070080, points=...)
>> at
>> /tmp/buildd/ros-cturtle-tabletop-object-perception-0.2.1/debian/ros-cturtle-tabletop-object-perception/opt/ros/cturtle/stacks/tabletop_object_perception/tabletop_object_detector/src/model_fitter.cpp:98
>> #13 0x080d4a3f in
>> tabletop_object_detector::DistanceFieldFitter::initializeFromMesh
>> (this=0x2070080, mesh=...)
>> at
>> /tmp/buildd/ros-cturtle-tabletop-object-perception-0.2.1/debian/ros-cturtle-tabletop-object-perception/opt/ros/cturtle/stacks/tabletop_object_perception/tabletop_object_detector/src/model_fitter.cpp:210
>> #14 0x080b3fbb in
>> tabletop_object_detector::ExhaustiveFitDetector<tabletop_object_detector::IterativeTranslationFitter>::loadDatabaseModels
>> (this=0xbfffec74,
>> model_set=...)
>> at
>> /tmp/buildd/ros-cturtle-tabletop-object-perception-0.2.1/debian/ros-cturt---Type
>> <return> to continue, or q <return> to quit---
>> le-tabletop-object-perception/opt/ros/cturtle/stacks/tabletop_object_perception/tabletop_object_detector/include/tabletop_object_detector/exhaustive_fit_detector.h:216
>> #15 0x080b9fc7 in TabletopNode (this=0xbfffeba4, nh=...)
>> at
>> /tmp/buildd/ros-cturtle-tabletop-object-perception-0.2.1/debian/ros-cturtle-tabletop-object-perception/opt/ros/cturtle/stacks/tabletop_object_perception/tabletop_object_detector/src/tabletop_node.cpp:231
>> #16 0x0807d457 in main (argc=1, argv=0xbffff004)
>> at
>> /tmp/buildd/ros-cturtle-tabletop-object-perception-0.2.1/debian/ros-cturtle-tabletop-object-perception/opt/ros/cturtle/stacks/tabletop_object_perception/tabletop_object_detector/src/tabletop_node.cpp:942
>>
>>
>> --
>> Ugo Cupcic | Shadow Robot Company |
>> ugo@shadowrobot.com <mailto:ugo@shadowrobot.com>
>> Software Engineer | 251 Liverpool Road |
>> need a Hand? | London N1 1LX | +44 20 7700 2487
>> http://www.shadowrobot.com/hand/ @shadowrobot
>>
>> _______________________________________________
>> ros-users mailing list
>> ros-users@code.ros.org <mailto:ros-users@code.ros.org>
>> https://code.ros.org/mailman/listinfo/ros-users
>>
>>
>>
>> _______________________________________________
>> ros-users mailing list
>> ros-users@code.ros.org <mailto:ros-users@code.ros.org>
>> https://code.ros.org/mailman/listinfo/ros-users
>
> --
> Ugo Cupcic | Shadow Robot Company | ugo@shadowrobot.com <mailto:ugo@shadowrobot.com>
> Software Engineer | 251 Liverpool Road | need a Hand? | London N1
> 1LX | +44 20 7700 2487 http://www.shadowrobot.com/hand/ @shadowrobot
>
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org <mailto:ros-users@code.ros.org>
> https://code.ros.org/mailman/listinfo/ros-users
>
>
>
>
> --
> Matei Ciocarlie
> Research Scientist
> Willow Garage Inc.
> 650-475-9780
>
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
(text/plain)