ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

I don't know about the 'best way', but two options I can come up with:

  • see if laser_filters in laser_pipeline can do something for you (probably distance based, ie: everything < 0.4m is the robot itself)
  • more complex: searching around it seems the Making collision maps from self-filtered laser data tutorial seems to do what you want, but then in 3D. After converting laser scans to PointClouds, it uses robot_self_filter amongst other things to filter out robot geometry based on the URDF and the robot's state (ie: TF). That will probably be more resource intensive, but could be much more precise

I'm not sure how much of an assumption on things being 3D there is in the last option, so that would be something to check.