PoseWithCovarianceStampedToGaussianPointCloud

Visualize geometry_msgs/PoseWithCovarianceStamped as gaussian pointcloud. Pointcloud is computed within a region of 3 sigma.

Subscribing Topics

  • ~input (geometry_msgs/PoseWithCovarianceStamped)

    Input pose

Publishing Topics

  • ~output (sensor_msgs/PointCloud2)

    Output pointcloud

Parameters

  • ~cut_plane (default: xy)

    You can choose a plane to compute gaussian distribution from xy, yz or zx.

  • ~sampling_num (default: 10)

    The number of sampling for each axis. The number of points will square of ~sampling_num.

  • ~normalize_method (default: normalize_area)

  • ~normalize_value (default: 1.0)

    You can choose normalize_area or normalize_height to normalize gaussian distribution. If you choo normalize_area, area of gaussian distribution will be ~normalize_value. If you choo normalize_height, the maximum height of gaussian distribution will be ~normalize_value.