mobilemanipulation / rgbd_ros_to_lcm Goto Github PK

sorry to bother you, I use your rgbd_ros_to_lcm package and Labelfusion to create 6d pose datasets, the collect lcm message process is first "roslaunch openni2_launch openni2.launch" second "rviz" third "rgbd_ros_to_lcm" and finally lcm-logger, but when I use the collected lcm message to 3d reconstruction the result is poor compare the collect lcm message process "./openni2-camera-lcm" and "lcm-logger".could you give me some advice on how to solve the problem, thanks for advance.

Sorry to bother you again and this issue is not associated with rgbd_ros_to_lcm, but I hope from your experience on LabelFusion might help. I am facing a new problem related to LabelFusion.

I am learning to use the LabelFusion pipeline to generate datasets from lcmlog file in the smaller sample dataset downloaded from the website. From my understanding, a file like above_table_pointcloud.vtp should be generated after completing the step "segment the pointcloud above the table". However, flowing the instructions here: https://github.com/RobotLocomotion/LabelFusion/blob/master/README.rst, I couldn't see any file like that generated. I also described this issue on github: RobotLocomotion/LabelFusion#28; But no answer up to now!

If I create my own data, is it necessary to have above_table_pointcloud.vtp? I attempted to execute gr.launchObjectAlignment('drill') without above_table_pointcloud.vtp in the folder, then errors occured.

Could you give me some suggestions?

Hello is it possible to use already existed RGB-D images and convert them to lcm?

Thank you!

Hi,
Thanks for your work!
I am trying to build the code and getting the flowing issue:

Any suggestion would be appreciated!

My system: ubuntu 16.04, ROS kinetic.

Hi, thank you for your ROS-LCM converter. I want to use it with realsense D415.
Firstly i installed the realsense2_camera as you adviced.and

roslaunch realsense2_camera rs_rgbd.launch

But after normal install rgbd_ros_to_lcm , when i run

roslaunch rgbd_ros_to_lcm lcm_republisher.launch

it produce error as follow:

it seems when the program run to line 162 of lcm_republisher.cpp, it can't step over it, and the process die
sensor_msgs::Image::ConstPtr rgb_msg = ros::topic::waitForMessage<sensor_msgs::Image>(rgb_topic, nh_);
My system: ubuntu 16.04, ROS kinetic.
Any suggestion would be appreciated!

I'm trying to subscribe to the native format of kinect 4 azure which is bgra8, I've made some changes to the code (mainly just editing bgr to bgra where it was possible). The topic runs for a while, at best I got to 12th frame and then I get an error:

 [lcm_republisher-1] process has died [pid 18753, exit code -11, cmd /home/ros/gabriel/rosToLcm/devel/lib/rgbd_ros_to_lcm/lcm_republisher __name:=lcm_republisher __log:=/home/ros/.ros/log/02590994-fed6-11e9-8f3f-00d8610e1481/lcm_republisher-1.log].
log file: /home/ros/.ros/log/02590994-fed6-11e9-8f3f-00d8610e1481/lcm_republisher-1*.log

When I try to open the log file it seems that is doesn't exist somehow. I'd be really glad if the issue could be resolved
Running ubuntu 18.04

my sensor is realsense D415 ，use together with Labelfusion https://github.com/RobotLocomotion/LabelFusion
i want to disable compress_rgb,compress_depth to get better quallity data.
here is my roslaunch file:
# input parameters
subscribe_point_cloud: false
#rgb_topic: /camera/rgb/image_rect_color
rgb_topic: /camera/color/image_raw
depth_topic: /camera/aligned_depth_to_color/image_raw

  #depth_topic: /camera/depth_registered/sw_registered/image_rect_raw
  cloud_topic: /camera/depth_registered/points

  # output parameters
  output_lcm_channel: "OPENNI_FRAME"
  compress_rgb: false
  compress_depth: false

  debug_print_statements: true

i recored success, but when i turn to labelfusion docker , run_trim this lcmlog file
error occur:

root@7735bcb889e1:~/labelfusion/data/log/test# run_trim
Found this many *lcmlogs:  1
Renaming to original_log.lcmlog
mv: 'original_log.lcmlog' and 'original_log.lcmlog' are the same file
Opening log player...
Opening viewer
OpenJDK 64-Bit Server VM warning: Using incremental CMS is deprecated and will likely be removed in a future release
LC: Disabling IPV6 support
LCM: Disabling IPV6 support
LCM: TTL set to zero, traffic will not leave localhost.
Exception: java.net.BindException: Address already in use (Bind failed)
Total seconds: 14.072087
==== LCM Warning ===
LCM detected that large packets are being received, but the kernel UDP
receive buffer is very small.  The possibility of dropping packets due to
insufficient buffer space is very high.

For more information, visit:
   http://lcm-proj.github.io/multicast_setup.html

Failed to get BotCamTrans for camera: OPENNI_FRAME_D
Failed to get coord_frame for camera: OPENNI_FRAME_D
unhandled image pixelformat 861030210 for camera OPENNI_FRAME_LEFT
WARNING: multisense_utils::unpack_multisense | image type not understood: 4
unhandled image pixelformat 861030210 for camera OPENNI_FRAME_LEFT
WARNING: multisense_utils::unpack_multisense | image type not understood: 4
unhandled image pixelformat 861030210 for camera OPENNI_FRAME_LEFT
WARNING: multisense_utils::unpack_multisense | image type not understood: 4
unhandled image pixelformat 861030210 for camera OPENNI_FRAME_LEFT
WARNING: multisense_utils::unpack_multisense | image type not understood: 4

i think this may not caused by rgbd_ros_to_lcm,but i was first time to know LCM,can you give me some tips ? thank you very much!

I have an issue similar to #2 where catkin_make fails with

/usr/bin/ld: cannot find -lvtkCommon
/usr/bin/ld: cannot find -lvtkFiltering
/usr/bin/ld: cannot find -lvtkImaging
/usr/bin/ld: cannot find -lvtkGraphics
/usr/bin/ld: cannot find -lvtkGenericFiltering
/usr/bin/ld: cannot find -lvtkIO
/usr/bin/ld: cannot find -lvtkRendering
/usr/bin/ld: cannot find -lvtkVolumeRendering
/usr/bin/ld: cannot find -lvtkHybrid
/usr/bin/ld: cannot find -lvtkWidgets
/usr/bin/ld: cannot find -lvtkParallel
/usr/bin/ld: cannot find -lvtkInfovis
/usr/bin/ld: cannot find -lvtkGeovis
/usr/bin/ld: cannot find -lvtkViews
/usr/bin/ld: cannot find -lvtkCharts
collect2: error: ld returned 1 exit status
rgbd_ros_to_lcm/CMakeFiles/lcm_republisher.dir/build.make:678: recipe for target '/root/catkin_ws/devel/lib/rgbd_ros_to_lcm/lcm_republisher' failed
make[2]: *** [/root/catkin_ws/devel/lib/rgbd_ros_to_lcm/lcm_republisher] Error 1
CMakeFiles/Makefile2:571: recipe for target 'rgbd_ros_to_lcm/CMakeFiles/lcm_republisher.dir/all' failed
make[1]: *** [rgbd_ros_to_lcm/CMakeFiles/lcm_republisher.dir/all] Error 2
Makefile:138: recipe for target 'all' failed
make: *** [all] Error 2
Invoking "make -j16 -l16" failed

I have ensured that libvtk6-dev and lvtk6 are installed, and I have even tried compiling vtk from source to no avail. I am attempting to install this in the current LabelFusion docker image, is there a specific version of LCM and PCL that must be installed for everything to work correctly? Should I give up with the docker image and just try to install everything on my host machine?

Just a quick look at the RealSense SR300 and Kinect 1 image encodings shows the following differences.

RealSense:

/camera/depth_registered/sw_registered/image_rect
encoding: 32FC1
/camera/depth_registered/sw_registered/image_rect_raw
encoding: 16UC1
/camera/rgb/image_raw
encoding: rgb8
/camera/rgb/image_rect_color
encoding: rgb8

Kinect:

/camera/depth_registered/sw_registered/image_rect
encoding: 32FC1
/camera/rgb/image_raw
encoding: bayer_grbg8
/camera/rgb/image_rect_color
encoding: bgr8

At minimum, should be able to handle input formats of 16UC1 and 32FC1 for depth, rgb8 and bgr8 for RGB images without any user input required.

Hi! Thanks a lot for your rgbd_ros_to_lcm.But I am very sorry that we failed to use rgbd_ros_to_lcm in labelfusion. Now, we can run your code in ros successfully ,but we can not find how the labelfusion subscribe the output