pcl_filters

ROS package to apply different filters to point clouds. It is used to provide two filtered point clouds to the rrt_planner (see launch/pcl_filters.launch). One is employed as sampling space, and the other one is used for traversability analysis of the 3D terrain.

General Parameters

• pointcloud_topic: ROS topic where the point cloud to be filtered is being published.

• output_topic: name of the ROS to topic where to publish the output point cloud filtered.

• odom_frame: TF frame where the odometry is being published.

• sensor_frame: TF frame of the sensor that is publishing the input point cloud.

• local_grid_radius: radius in meters of the local area considered around the robot where the global point cloud must be cropped.

• filter_ceiling: boolean to indicate whether the ceiling and the walls should be removed (partially) from the point cloud.

• show_normal_arrows: boolean to indicate if the normal vectors of the surface should be published to be visualized in RViz.

• pith_max_inclination: surfaces that exceed this maximum inclination of the pitch will be removed. (Only applicable if filter_ceiling is true).

• roll_max_inclination: surfaces that exceed this maximum inclination of the roll will be removed. (Only applicable if filter_ceiling is true).

• max_roughness: surfaces that exceed this maximum terrain roughness will be removed. (Only applicable if filter_ceiling is true).

PCL filters Parameters

These are provided through a yaml file for sake of clarity (see examples launch/rrt_filter_params.yaml and launch/3dfeatures_params.yaml).

Possible filters:

• StatisticalOutlierFilter

• PassThroughFilter

• CropBoxFilter

• VoxelGridFilter

• VoxelGridCovarianceFilter

Dependences

• PCL is required in its version 1.9.1.

The package is a work in progress used in research prototyping. Pull requests and/or issues are highly encouraged.