softkinetic_camera

General description

This package implements a node that runs an interactive gesture camera (softkinetic) in ROS

To use this package in ROS, the DepthSenseSDK for linux needs to be downloaded. The download can be found here: http://www.softkinetic.com/fr-be/support/download.aspx?EntryId=517 To download, you will need to create a free account.

Parameters

camera_link (string, default: “/softkinetic_link”)
The frame ID of the camera.

confidence_threshold (int, default: 150)
Confidence threshold for DepthNode configuration. Sensor noise is filtered by increasing the threshold. Threshold needs to be within [0, 32767].

Published Topics

depth/points (sensor_msgs::PointCloud2)
Publishes point clouds detected by the camera.

rgb_data (sensor_msgs::Image)
Publishes rgb-images in bgr8 encoding.

…

Add a camera to your robot model

Here we explain how to import a softkinetic camera into your robot xacro file. We have only included a descriptor for the Creative Senz3D camera, but adapt it to any Softkinetic camera should be trivial. You just need to import the xacro file describing the camera macro:

  <xacro:include filename="$(find softkinetic_camera)/urdf/senz3d.urdf.xacro"/>

And include it somewhere in your robot description file, e.g.:

  <sensor_senz3d parent="base_link" name="senz3d_camera">
    <origin xyz="0.05 0.0 0.3" rpy="0.0 0.0 0.0"/>
  </sensor_senz3d>

Changelog for package softkinetic_camera

0.6.2 (2016-06-23)

0.6.1 (2016-06-21)

• 0.6.1
• added CHANGELOG
• consistent version nummber
• Revert "Point cloud reformation to kinect-like "
• demo launch file modified according to frame_id change and standardize aspect ratio
• compile dummy node if the depthsense sdk is not found (used for travis compilation)
• remove non-rgb pointcloud
• point cloud frame_id fixed to rgb_optical_frame
• Merge pull request #55 from knorth55/serial-cfg-fix fix dynamic_reconfigure config
• add time stamp in PointCloud ROS msg
• fix dynamic_reconfigure config
• Merge pull request #54 from knorth55/add-keep-organized add setKeepOrganized to get organized output
• Merge pull request #53 from knorth55/filter-default-cfg-fixed Fix use_filter default value True -> False
• Merge pull request #52 from knorth55/image-format-fixed Fix image format WXGA -> WXGA_H
• add use_serial in cfg
• add use_serial and serial rosparam
• add setKeepOrganized to get organized output
• fix use_filter default value True -> False
• fix image format WXGA -> WXGA_H
• Remove useles variables
• Get some performance improvement by not processing 3D points for saturated pixels (as they are anyway all-zero points). Also enforce calling any other filter before radious outlier removal so we get rid of the all-zero points and it doesn't take ages (I have tweaked default parameters accordingly).
• Fix fustrum filter camera correction acording to the projection in indigo (that turned to be wrong! see issue #46). Also, fustrum filter only works with fov < 180, so I set max values to 180 (and also set more realistic bounds for close/far planes)
• Code simplification
• Merge branch 'pal-robotics-forks-hydro_dev' into merge_pal Intermedate step before merging into indigo. I retain from PAL PR the dynamir reconfigure, the yaml configuration and the two additional filters (though frustum culling is not working!)
• disable not used outputs on depth node (doesn't improve performance, anyway)
• Replace ProjectionHelper by the UVmap; seems to be a bit faster and for sure makes the code a bit cleaner
• Fix the collision box orientation. Also set realistic values for inertia , calculated with the solid cuboid formula from here: http://en.wikipedia.org/wiki/List_of_moments_of_inertia
• Convert softkinetic vertices into a kinect-like coordinates pointcloud. Required by 83b8a8f2cfdcdc0441e4b433ad86cb16f378e54b
• Replicate kinect tf scheme.
• Merge branch 'hydro_dev' of https://github.com/pal-robotics/softkinetic into hydro_dev
• fix variable name
• fixes minor warnings and cosmetics
• adds range_max param
  • reprojects x, y, z = range_max when no reading
  • uses i=j=0 instead of i=j=1 (last 2 rows were discarded)
• Merge remote-tracking branch 'upstream/hydro_dev' into hydro_dev Conflicts: softkinetic_camera/CMakeLists.txt
• adds frustrum culling filter
• Issue #25: Color image is wrongly formed because the step field is not set
• Pull request #30: a complete driver revision of the driver Also puts indigo branch up to date with hydro, so PR #33 is unnecessary
• dynamic reconfigure parameter not taking effect confidence_threshold update through dynamic reconfigure was not being applied
• Merge branch 'hydro_dev' of https://github.com/pal-robotics/softkinetic into hydro_dev
• fix reading of /camera_link parameter a private NodeHandle is used to retrieve "camera_link" parameter. However, the parameter name was set to "/camera_link" so that it was being looked for in the root namespace and therefore, not found.
• fix setupCameraInfo to use cam_info of the method's arg
• Replicate PR #32 to speed up both color and depth images reading
• use memcpy to copy color image instead of nested for-loops
• renames voxel_grid_size and min_neighbours params
• base_rgbd_camera_optical_frame on static_transform
• drops outdated opencv2 dependency for indigo
• adds dynamic reconfigure server
• updates indentation
• adds passthrough filter params
• adds passthrough filter
• Merge pull request

File truncated at 100 lines see the full file