Drone SLAM + YOLO integration

This repo contains files that every robotics UAV engineer would need :

This repository integrates PX4, ROS 2 Jazzy, and Gazebo Harmonic for drone navigation and control using SLAM.
It provides launch files, configuration, and example workflows to visualize LiDAR scans, build maps, and control a drone in simulation.

Features:

• drone_slam: start the simulation, bridges, SLAM node, and visualization in RViz2 using 2D lidar .

• drone_slam_cam: Uses a RGBD camera mounted on the drone for SLAM and point cloud extraction.

• px4_rl_project: this folder contains a DQN approach for drone hovering and altitude maintenance

• yolo_ros repo could be exploited in future usage to do object detection and S&R tasks-- commands to use yolo_ros will be found in the folder.

• An odometry converter script (odom_converter.py) to transform PX4 vehicle odometry to ROS2-compatible format (needed in Rviz to establish a proper TF tree).

• Keyboard control node which you could find in drone_slam and drone_slam_cam

Note: This project uses a Python-based odometry converter by default (via the launch file), but a C++ version (odom_converter.cpp) is also provided for alternative use. Note: This repo is for having a good setup in place, future control integration and Nav2 or SkyPilot deployment will follow later !!

Needed Repos

1. Install prerequisites

sudo apt update
sudo apt install python3-colcon-common-extensions git

2. Create a workspace

mkdir -p ~/px4_ros2_ws/src
cd ~/px4_ros2_ws/src

3. Clone the repos

# px4_msgs contains ROS2 message definitions
git clone https://github.com/PX4/px4_msgs.git

# px4_ros_com contains the microRTPS bridge & ROS2 interfaces
git clone https://github.com/PX4/px4_ros_com.git

4. Build the workspace

From the workspace root:

cd ~/px4_ros2_ws
colcon build

5. Source the workspace

source install/setup.bash

If you want it permanent:

echo "source ~/px4_ros2_ws/install/setup.bash" >> ~/.bashrc

6. Verify installation

Check if the PX4 messages are available:

ros2 interface show px4_msgs/msg/VehicleOdometry

If it prints the message definition → ✅ you installed px4_msgs. Also check:

ros2 pkg list | grep px4

PX4-ROS2-SLAM-Control/
 ├── drone_slam/            # SLAM with Lidar 2D, TD3 control and keyboard control 
 ├── drone_slam_cam/        # SLAM with point cloud and RGBD
 ├── px4_rl_project/         # Reinfocement learning Control for drone hovering
 ├── yolo_ros/                  # Commands for yolo_ros that you will need
 └── README.md

Known Issues

• Setting fixed frame to map requires SLAM to be active.

• odom frame may appear rotated vs base_link.

Contributions

⭐⭐ If you found these projects interesting don't forget that star ⭐⭐ !!