PX4 + ROS 2 Offboard Control Framework

An engineering-grade framework for PX4 Offboard Control using ROS 2 Humble. This project implements a high-performance trajectory tracking node utilizing Micro-XRCE-DDS for low-latency communication and Velocity Feed-Forward logic to minimize tracking error during dynamic maneuvers.

🏗 System Architecture

The system is designed to bridge high-level path planning (ROS 2) with low-level flight control (PX4) via the uORB middleware.

• Middleware: Micro-XRCE-DDS (UDPv4)
• Control Strategy: Position Control + Velocity Feed-Forward (to eliminate phase lag)
• Simulation: Gazebo (Garden/Harmonic) or Classic
• Visualization: RViz2 & PlotJuggler (Pre-configured)

🚀 Installation & Setup

This repository uses Git Submodules to manage dependencies like PX4-Autopilot and px4_msgs.

1. Clone the Repository

You must clone with the --recursive flag to automatically download the required PX4 submodules.

git clone --recursive [https://github.com/GRawhideMart/px4_offboard_tracking.git](https://github.com/GRawhideMart/px4_offboard_tracking.git)
cd px4_offboard_tracking

If you forgot to clone recursively, run: git submodule update --init --recursive

2. Build the ROS 2 Workspace

Ensure you have a working ROS 2 Humble environment.

cd ros2_ws

# Source ROS 2 Humble
source /opt/ros/humble/setup.bash

# Build the workspace (including px4_msgs and the control node)
colcon build

# Source the overlay
source install/setup.bash

3. Build & Run PX4 SITL (Simulation)

In a separate terminal, build and launch the PX4 Software In The Loop (SITL) simulation.

cd PX4-Autopilot
make px4_sitl gz_x500

Wait until the drone spawns in Gazebo.

---

🚁 How to Run

Once the simulation is running, launch the entire control stack with a single command:

# In a new terminal (inside the project root)
source ros2_ws/install/setup.bash

# Launch the Agent, Micro-DDS, RViz2, and PlotJuggler
ros2 launch px4_trajectory_tracking system_integration.launch.py

Expected Behavior

1. MicroXRCE-DDS Agent connects to the simulator on UDP port 8888.
2. The drone performs a pre-flight check and arms automatically.
3. The drone takes off to an altitude of -5.0m.
4. It begins tracking a circular trajectory with feed-forward velocity inputs.

---

🧠 Control Logic: The "Feed-Forward" Approach

A standard position controller () often suffers from steady-state error (lag) when tracking moving targets. To solve this, this node computes the analytical derivative of the trajectory:

By sending both Position and Velocity setpoints to PX4, the internal flight controller uses the velocity vector as a feed-forward term, reducing phase lag to near zero.

# Simplified Logic
target_x = RADIUS * cos(omega * t)
target_vx = -RADIUS * omega * sin(omega * t)  # Feed-Forward Term

msg.position = [target_x, target_y, altitude]
msg.velocity = [target_vx, target_vy, 0.0]

🛠 Dependencies

• ROS 2 Humble (Desktop Full recommended)
• PX4 Autopilot (v1.14+)
• Micro-XRCE-DDS Agent
• Python 3.10+ (rclpy, numpy)

📄 License

This project is open-source and available under the MIT License.