UAV Tracking with Solid-State Lidars: Dynamic Multi-Frequency Scan Integration

Project Page   •   Paper   •   Contact Us

This is the code implementation for the paper UAV Tracking with Solid-State Lidars: Dynamic Multi-Frequency Scan Integration.

The code has been tested on Ubuntu 20.04 with ROS Noetic

• PCL (Point Cloud Library)
• Eigen3
• Boost
• ROS Noetic
• Livox_ros_driver: Follow livox_ros_driver Installation

Note: The drone detection requires the point cloud to be in PointCloud2 format while the tracking uses the Livox CustomMsg data type. For the conversion use this repository and follow the instructions. If you use a different repository, change the value for the self.lidar_sub variable in the livox_to_img.py file accordingly.

cd ~/catkin_ws/src

git clone https://github.com/IacopomC/dynamic_scan_tracking

cd ~/catkin_ws

catkin build dynamic_scan_tracking

source devel/setup.bash

# Terminal 1: Launch the tracking system
roslaunch dynamic_scan_tracking dynamic_scan_tracking.launch

# Terminal 2: Start your LiDAR driver (example for Livox)
roslaunch livox_ros_driver livox_lidar_msg.launch

# The system will start tracking from the configured initial position

Edit config/dynamic_scan_tracking.yaml:

kalman_filter:
  motion_model:
    type: "adaptive"  # Options: "cv", "ctrv", "adaptive"

• "adaptive": Automatically handles both straight flight and turns
• "cv": Best for straight-line motion, fastest computation
• "ctrv": Best for circular motion with known turn rate

• Tracking accuracy: Adjust search_radius (0.1-0.5m)
• Motion agility: Adjust process_noise.noise_a (1.0-5.0)
• LiDAR noise: Adjust measurement_noise values (0.05-0.2m)

See TECHNICAL_DOCUMENTATION.md for detailed parameter explanations.

• /dynamic_scan_tracking/object_pose: Current target pose
• /dynamic_scan_tracking/object_velocity: Current target velocity
• /dynamic_scan_tracking/target_pcl: Tracked point cloud (visualization)

• TECHNICAL_DOCUMENTATION.md: Comprehensive technical guide covering system architecture, motion models, configuration, and troubleshooting
• config/dynamic_scan_tracking.yaml: Configuration file with detailed parameter explanations

If you use this code for any academic work, please cite the following publication:

@inproceedings{catalano2023uav,
  title={Uav tracking with solid-state lidars: dynamic multi-frequency scan integration},
  author={Catalano, Iacopo and Sier, Ha and Yu, Xianjia and Westerlund, Tomi and Queralta, Jorge Pena},
  booktitle={2023 21st International Conference on Advanced Robotics (ICAR)},
  pages={417--424},
  year={2023},
  organization={IEEE}
}