Skip to content

README.md

Latest commit

 

History

History
106 lines (71 loc) · 3.72 KB

README.md

File metadata and controls

106 lines (71 loc) · 3.72 KB

🛸 Multi-Drone PX4 SITL Simulation with Realsense Support

This package enables multi-drone Software-in-the-Loop (SITL) simulations using PX4 and Gazebo Classic with:

  • ✅ Support for multiple IRIS quadrotors
  • 📸 Optional integration of Intel Realsense D435i (simulated)
  • 🌍 A supervisor node for managing transformations and common coordinate frames

🧪 If it works here, it works in the real world. This simulation is designed to closely mimic real flight conditions, especially for multi-agent coordination and localization setups.

🚀 Launch Files

This package provides a set of ROS launch files to start different simulation configurations:

Launch File Description
iris_sim.launch Launches a multi-drone simulation without cameras. Ideal for performance testing or formation flight.
uvify_sim.launch Launches a multi-drone simulation with Realsense D435i sensors for each drone. Use for perception, SLAM, and vision-based tasks.

More launch files will be added soon.

➕ Spawning Multiple Drones

To simulate multiple drones:

  1. Open your desired launch file (e.g., uvify_sim.launch).

  2. Locate the following XML group:

    <group ns="uav0">
  ...
</group>

  3. Copy and paste the group block for each additional drone.

  4. Update:

    • The ns attribute (uav1, uav2, etc.)
    • All ID, namespace, or related identifiers inside the block.

Caution

Ensure unique starting positions for each drone to avoid collisions in simulation! Update the initial <x>, <y>, <z> coordinates for every robot instance.

🧭 The Supervisor Node

The supervisor node plays a critical role in multi-drone coordination:

  • Establishes a common world frame across all drones.

  • Manages frame transforms:

    map → uavX/local_origin → uavX/base_link

  • Publishes relative odometry for each UAV with respect to the global fixed frame.

Note

In real-world deployments, this is analogous to the shared GPS frame every drone uses. In simulation, we mimic this using fixed transforms and simulated position feedback.

🌐 Geodetic Reference & Local Origin

The supervisor defines each drone’s local_origin using geodetic coordinates (latitude/longitude). This is converted into the simulation frame using the pyproj library.

📦 Install pyproj:

pip install pyproj

Warning

Accurate GPS data or a high-quality EKF2 configuration is essential for real-world applications. Poor position estimates can lead to large transform drift and localization errors.

🛠 Configuration

To configure which drones are spawned and tracked:

  1. Open the file:

    config/config.yaml

  2. Set the list of UAV IDs you want to include in the simulation or real-world setup. For example:

    drone_ids: [0, 1, 2]

Tip

These IDs should match the namespaces used in your launch file (uav0, uav1, ...).

🔧 Coming Soon

We're actively improving this package. Upcoming features include:

  • 🔄 Dynamic UAV spawner
  • 📡 MAVLink telemetry multiplexing
  • 🗺️ Map server integration
  • 🎥 Enhanced Realsense simulation realism

🤝 Contributing

Have ideas or found a bug? Contributions are welcome!