LiDAR cone detection algorithm

Requirements

You should have:

• Ubuntu 18.04 (Should work with latest versions).
• ROS melodic.

⚠️ Tested with Velodyne's VLP-16 LiDAR.

How to start using this repo?

• The root folder of this repository is supposed to be your catkin workspace.

Installation and Setup

1. Clone the repository and either source the new folder called "lidar" or change the name to your existing catkin workspace directory name.
2. Install the necessary ros packages by cloning them into the src folder:

• https://github.com/catkin/catkin_simple.git
• https://github.com/ethz-asl/glog_catkin.git

3. Install glog: sudo apt-get install libgoogle-glog-dev
4. Install velodyne depencencies for ROS: sudo apt-get install ros-melodic-velodyne
5. Install ros-numpy: sudo apt-get install ros-melodic-ros-numpy
6. Install scikit-learn: pip install -U scikit-learn
7. Build with catkin_make
8. Source the setup file by running: source ~/<catkin_workspace_name>/devel/source.bash. If you are using zsh instead of bash run: source ~/<catkin_workspace_name/devel/source.zsh

Launch instructions

To launch the whole environment run: roslaunch lidar_nodes lidar.launch.

Getting up and running with your own point cloud source should be as simple as:

1. Change the input_topic parameter in segmentation.launch to your topic.(Already configured, change only if you want to run your custom setup)
2. Adjust the sensor_height parameter in segmentation_params.yaml to the height where the sensor is mounted on your robot (e.g. KITTI Velodyne: 1.8m)

Visualize the data

1. Open RVIZ with the frame set to velodyne: rosrun rviz rviz -f velodyne
2. Click the add button and choose the topic you want to listen to:

• /velodyne_points: raw points captured by the lidar.
• /prefiltered_points: Remaining points after filters(depth, azimuth, etc.)
• /ground_cloud: Ground points.
• /obstacle_cloud: Cone pointcloud.
• /visualization_marker: Detected cones.

Ground segmentation parameter description

Parameters are set in linefit_ground_segmentation_ros/launch/segmentation_params.yaml

This algorithm works on the assumption that you known the height of the sensor above ground. Therefore, you have to adjust the sensor_height to your robot specifications, otherwise, it will not work.

The default parameters should work on the KITTI dataset.

Ground Condition

• sensor_height Sensor height above ground.
• max_dist_to_line maximum vertical distance of point to line to be considered ground.
• max_slope Maximum slope of a line.
• max_fit_error Maximum error a point is allowed to have in a line fit.
• max_start_height Maximum height difference between new point and estimated ground height to start a new line.
• long_threshold Distance after which the max_height condition is applied.
• max_height Maximum height difference between line points when they are farther apart than long_threshold.
• line_search_angle How far to search in angular direction to find a line. A higher angle helps fill "holes" in the ground segmentation.

Segmentation

• r_min Distance at which segmentation starts.
• r_max Distance at which segmentation ends.
• n_bins Number of radial bins.
• n_segments Number of angular segments.

Other

• n_threads Number of threads to use.
• latch Latch output point clouds in ROS node.
• visualize Visualize the segmentation result. ⚠️ only for development and debugging, don't use during operation.