ar_bounding_box detects a set of marker patterns and creates a coordinate system and a bounding box around them. The point cloud inside this bounding box is published and can be used by re_object_recorder to create object models.
This package is based on Michael Ferguson's ar_kinect package.
This package is part of RoboEarth (EU FP7, grant 248942).
- Author: Andreas Koch
- License: GPL
- Source: svn https://ipvs.informatik.uni-stuttgart.de/roboearth/repos/public/tags/latest
ar_kinectEstimates a bounding box around a marker pattern using the Kinect camera and ARToolkit and publishes the contents of this bounding box as point cloud.
Published Topicsresulting_pcl (sensor_msgs/PointCloud2)
- The resulting point cloud, limited to the bounding box
- A point cloud with 4 points, describing the bounding box
Parametersboxsize (double, default: 0.135)
- Edge length of the bounding box
- The maximum allowed error in detecting the markers. If you get too many "Distance too large" errors, increase this value. Please note that the model quality will decrease with larger values.
Provided tf Transforms"/camera_rgb_optical_frame" → "/PATTERN_BASE"
- The transformation from the camera to the object coordinate system