pointcloud_crop_box

A ROS 2 package for filtering 3D point clouds using multiple axis-aligned cropping boxes. It subscribes to a sensor_msgs/msg/PointCloud2 or sensor_msgs/msg/LaserScan, transforms it into a target frame, filters it using configurable crop boxes, and publishes both the filtered cloud and visualization markers for the crop boxes.

---

Features

• Subscribes to raw point cloud data (e.g., from a LiDAR) or laser scan data
• Filters the cloud using multiple configurable 3D crop boxes
• Supports TF2 transformation to a target frame
• Option to invert the box filtering (i.e., remove inside instead of outside)
• Publishes the filtered cloud and visualization markers for the crop boxes

---

ROS 2 Node

pointcloud_crop_box_node

Subscribes

• <input_topic> (sensor_msgs/msg/PointCloud2 or sensor_msgs/msg/LaserScan) Input topic containing raw data (default: /points_raw)

Publishes

• <output_topic> (sensor_msgs/msg/PointCloud2) Filtered point cloud after box cropping (default: /points_filtered)

• ~/filter_crop_boxes (visualization_msgs/msg/MarkerArray) Visualization markers for all configured crop boxes, for visualization and debugging

---

Parameters

• input_topic [string, default: "/points_raw"]: Topic to subscribe for input PointCloud2 data.
• output_topic [string, default: "/points_filtered"]: Topic to publish filtered PointCloud2 data.
• target_frame [string, default: "base_link"]: Target TF frame to transform the point cloud into.
• crop_boxes_names [string_array, default: ["robot_box"]]: List of crop box names to apply sequentially.
• crop_boxes [map]: Configuration for multiple crop boxes, with each crop box having the following parameters:
  • min_x [double, default: -1.0]: Minimum X boundary of the crop box.
  • max_x [double, default: 1.0]: Maximum X boundary of the crop box.
  • min_y [double, default: -1.0]: Minimum Y boundary of the crop box.
  • max_y [double, default: 1.0]: Maximum Y boundary of the crop box.
  • min_z [double, default: -1.0]: Minimum Z boundary of the crop box.
  • max_z [double, default: 1.0]: Maximum Z boundary of the crop box.
• message_type [string, default: "pointcloud"]: type of source data to crop, 'pointcloud' or 'laserscan'.
• negative [bool, default: false]: If true, keeps points outside the crop boxes instead of inside.
• visualize_crop_boxes [bool, default: true]: Whether to publish visualization markers for the crop boxes.

---

Configuration

Multiple Crop Boxes

This package supports multiple crop boxes that are applied sequentially to filter the point cloud. You can configure multiple crop boxes in the parameter file:

/**:
  pointcloud_crop_box:
    ros__parameters:
      target_frame: base_link
      crop_boxes_names: ["robot_box", "robot_attachment_1", "robot_attachment_2"]
      crop_boxes:
        robot_box:
          min_x: -0.5
          max_x: 0.5
          min_y: -0.5
          max_y: 0.5
          min_z: -0.1
          max_z: 0.3
        robot_attachment_1:
          min_x: -2.0
          max_x: 2.0
          min_y: -1.0
          max_y: 1.0
          min_z: 0.5
          max_z: 2.0
        robot_attachment_2:
          min_x: -5.0
          max_x: 5.0
          min_y: -5.0
          max_y: 5.0
          min_z: 2.5
          max_z: 5.0
      negative: false
      message_type: pointcloud
      visualize_crop_boxes: true

---

Example Launch

To show the result of the filtering you can follow our autonomy example in husarion-ugv-autonomy to get measurements of pointcloud.

Build from Source

Create Workspace

mkdir ~/husarion_ws
cd ~/husarion_ws
git clone https://github.com/husarion/pointcloud_crop_box.git src/pointcloud_crop_box

Build

sudo rosdep init
rosdep update --rosdistro $ROS_DISTRO
rosdep install --from-paths src -y -i

source /opt/ros/$ROS_DISTRO/setup.bash
colcon build --symlink-install --packages-up-to pointcloud_crop_box --cmake-args -DCMAKE_BUILD_TYPE=Release

Run configured launch:

source install/setup.bash
ros2 launch  pointcloud_crop_box pointcloud_crop_box_launch.py