Triggered camera / multicamera plugins #687
Conversation
|
I filed an issue with gazebo because the first triggered image looks different (and wrong) compared to subsequent images: |
| double _hack_baseline) | ||
| { | ||
| GazeboRosCameraUtils::Load(_parent, _sdf, _camera_name_suffix, _hack_baseline); | ||
|
|
There was a problem hiding this comment.
add another this->SetCameraEnabled(false); here in this overloaded Load function?
| ros::spinOnce(); | ||
| ros::Duration(0.1).sleep(); | ||
| } | ||
| EXPECT_EQ(images_received_, 2); |
There was a problem hiding this comment.
is there logic to filter out mgs close together? I'm getting an error:
[gazebo_plugins.rosunit-triggered_camera/cameraSubscribeTest][FAILURE]----------
/home/osrf/code/ian/gz_ws/src/gazebo_ros_pkgs/gazebo_plugins/test/camera/triggered_camera.cpp:103
Value of: 2
Expected: images_received_
Which is: 3
but the tests are passing on jenkins..
There was a problem hiding this comment.
hm...it is kind of a race condition, but it was pretty consistent on my machine
are you consistently getting this result?
There was a problem hiding this comment.
It was pretty consistent when I was testing it before but I just ran the test 10 more times and it only failed once so likely a race condition.
There was a problem hiding this comment.
an alternative is to try to change the behavior so that each trigger signal leads to a new image. this would require a slight refactoring (use int instead of bool in a few places) but would be more predictable
There was a problem hiding this comment.
I think I've fixed the race condition in cb9f299. Now the plugin uses an int to keep track of the number of triggers received from the topic.
scpeters
force-pushed
the
triggered_cameras_kinetic
branch
from
00a0a97
to
6752f51
Compare
|
I just rebased on kinetic-devel in order to get the renamed distortion tests. |
scpeters
force-pushed
the
triggered_cameras_kinetic
branch
from
6752f51
to
7875cbd
Compare
|
mmm the new builds are failing because |
|
@scpeters would you mind fixing the conflicts when you have some free time? Thanks. |
scpeters
force-pushed
the
triggered_cameras_kinetic
branch
from
cb9f299
to
9cd1ae5
Compare
|
@j-rivero sorry for the delay; I just resolved the conflicts |
|
I tested locally with gazebo9 and the tests pass except for the barrel distortion test (as expected due to #681 ) |
maybe we aren't declaring an explicit dependency on it; let me check |
|
Someone else had the |
|
@ros-pull-request-builder retest this please |
1 similar comment
|
@ros-pull-request-builder retest this please |
|
@ros-pull-request-builder retest this please |
|
The builds are now unstable, with just the distortion camera test failures |
|
just tested everything again, works as expected. |
|
I've added the missing entry to |
* adds triggered cameras and multicameras * fix threaded event connection * use correct timestamp for images * remove compiler directives for old gazebo versions * update copyright dates and remove copied comments * test for triggered_camera * expect cameras don't subscribe to trigger topic * subscribe to trigger if CanTrigger is true * fix untriggered first image * fix race condition * change triggered camera test name * fix test * fix 16bit test name
This adds a plugin for triggered cameras and multicameras. These sensors do not publish unless triggered. They have an additional topic (default name
image_trigger) that subscribes tostd_msgs/Emptymessages and will publish a single update after being triggered.Its maximum update rate is currently set by the
<update_rate>sdf tag in the<sensor>block. Under the hood. If you publish two trigger messages before the camera has a chance to publish, it will only publish once, since aboolis used to keep track of the trigger status. The test reflects this behavior.