DBSCAN_ROS

A high-performance Adaptive DBSCAN clustering ROS node that segments point clouds into clusters for downstream perception tasks. It processes a filtered LiDAR/RADAR point cloud, runs adaptive DBSCAN clustering, and outputs:

1. Clustered Point Clouds (via autoware_perception_msgs::DynamicObjectWithFeatureArray)
2. Markers visualizing bounding boxes (via visualization_msgs::MarkerArray)
3. Noise Points (points not belonging to any cluster)

This node is especially useful for 3D LiDAR-based perception in autonomous driving or robotics systems, where you need to:

• Identify objects in point cloud data,
• Publish bounding boxes for visualization,
• And keep track of any points that do not belong to relevant objects.

Table of Contents

1. Features
2. Dependencies
3. Installation and Build Instructions
4. Usage
5. Node Parameters
6. Key Parts of the Code
7. Performance
8. Contribution
9. License

Dependencies

This package depends on common ROS and perception libraries:

• ROS (tested with ROS melodic/noetic)
• PCL (Point Cloud Library)
• sensor_msgs
• visualization_msgs
• autoware_perception_msgs (from Autoware, or your custom message definitions)

Ensure you have these dependencies in your package.xml:

<depend>roscpp</depend>
<depend>sensor_msgs</depend>
<depend>visualization_msgs</depend>
<depend>pcl_conversions</depend>
<depend>pcl_ros</depend>
<depend>autoware_perception_msgs</depend>

Installation and Build Instructions

cd ~/catkin_ws/src
git clone https://github.com/your-username/dbscan_ros.git
cd ~/catkin_ws
catkin_make
source devel/setup.bash

Usage

1. Start ROS Master:

   roscore

2. Launch LiDAR input (e.g., bag file, driver, or simulation)
3. Run dbscan_ros:

   rosrun dbscan_ros dbscan_ros_node

4. Visualize outputs in rviz with topics:
   • /dbscan_clustering/marker (bounding boxes)
   • /dbscan_clustering/noise_points (outliers)

Node Parameters

Parameter Name	Type	Default Value	Description
topic_input	string	/rslidar_points_front/ROI/no_ground	Input point cloud topic
topic_output	string	/dbscan_clustering/clusters	Output topic for clusters
obj_width_thre	double	0.15	Approximate object width threshold
obj_height_thre	double	0.15	Approximate object height threshold
downsample_resolution	double	0.03	Resolution for downsampling
lidar_horizon_resolution	double	0.2	LiDAR horizontal resolution (degrees)
lidar_vertical_resolution	double	1.0	LiDAR vertical resolution (degrees)

Key Parts of the Code

1. Parameter Initialization

nh_private.param<std::string>("topic_input", topic_input_, "/rslidar_points_front/ROI/no_ground");
nh_private.param<double>("obj_width_thre", obj_width_thre_, 0.15);

2. Subscriber and Publishers

pc_sub = nh.subscribe(topic_input_, 1, &dbscan_ros::lidar_callback, this);
cluster_marker_pub = nh.advertise<visualization_msgs::MarkerArray>("/dbscan_clustering/marker", 10);

3. The Clustering Process

AdaptiveDBSCANCluster<pcl::PointXYZ> db;
db.setInputCloud(ground_remo);
db.setSearchMethod(tree);
db.setClusteringParams(obj_width_thre_, num_points_per_line_, max_num_lines_, eps_with_distance_factor_, num_line_with_distance_factor_);
db.setMinClusterSize(8);
db.setMaxClusterSize(20000);
db.extract(indices);

4. Bounding Box and Center Calculation

for (pcl::PointIndices ind : indices) {
    // Compute min, max, center, and create markers
}

5. Noise Points Extraction

pcl::PointIndices::Ptr all_cluster_indices_ptr(new pcl::PointIndices(all_cluster_indices));
ext.setIndices(all_cluster_indices_ptr);
pcl::PointCloud<pcl::PointXYZ> noise_cloud;
ext.setNegative(true);
ext.filter(noise_cloud);

6. Performance Measurement

auto start = std::chrono::steady_clock::now();
auto end = std::chrono::steady_clock::now();
auto duration_ = std::chrono::duration_cast<std::chrono::milliseconds>(end - start).count();
if (duration_ >= 10 ) std::cout << "time took is, " << duration_ << std::endl;

Performance

• Real-time or near real-time performance depending on:
  • CPU capabilities
  • Density of point cloud
  • Complexity of scene
• Adaptive factors keep clusters well-defined even with different LiDAR resolutions.

Contribution

Contributions are welcome! To contribute:

1. Fork the repository.
2. Create a feature branch (git checkout -b feature-name).
3. Commit your changes (git commit -m 'Add new feature').
4. Push the branch (git push origin feature-name).
5. Open a pull request.

License

This code is provided under MIT. Please refer to the LICENSE file for details.

Thank You

Enjoy using dbscan_ros for your LiDAR clustering needs! If you run into issues, feel free to open an issue or contribute via pull requests.

Happy Clustering!