PX4 SITL Multi-Vehicle Launcher

Launch one or more PX4 SITL vehicles in Gazebo from a single scenario script.

Setup

chmod +x launch_*.py

Add this to your ~/.bashrc (then open a new shell):

export PX4_Autopilot_Dir=~/PX4-Autopilot

Running

./launch_multi_atmos.py

The launcher runs in the foreground and prefixes each vehicle's output with its name. Press Ctrl-C to stop every vehicle and its Gazebo process group. Scenario state files are stored under ~/.cache/px4-sitl/.

Configuring

Each scenario script has a small CONFIG block at the top. Edit it to change the world, the state session name, or the vehicle list (names, models, poses):

WORLD   = "kthspacelab"
SESSION = "atmos"

VEHICLES = [
    Vehicle(name="snap",    model="gz_atmos", pose=(1, 0, 0.2)),
    Vehicle(name="crackle", model="gz_atmos", pose=(2, 0, 0.2)),
]

Vehicle.name is used for the DDS namespace, live output prefix, and status output. Gazebo assigns its own model names such as atmos_0 and atmos_1; this is intentional because setting PX4_GZ_MODEL_NAME makes PX4 attach to an existing Gazebo model instead of spawning one.

To add a new model, edit px4_sitl_launcher.py and add it to both MODEL_AUTOSTART and MODEL_BUILD.

Managing a Scenario

./launch_multi_atmos.py --detach   # launch and return to the shell
./launch_multi_atmos.py --status   # show vehicle PIDs and states
./launch_multi_atmos.py --stop    # stop only this scenario

Detached scenarios are still owned by their individual process groups, so stopping one scenario does not kill unrelated PX4 or Gazebo processes.

Notes

ROS 2 integration. In a separate terminal, run the XRCE-DDS agent:

micro-xrce-dds-agent udp4 -p 8888

Each vehicle's topics then appear under its name, e.g. /snap/fmu/out/....

Troubleshooting

PX4 binary not found at .../build/<target>/bin/px4 — build the target it asks for:

cd $PX4_Autopilot_Dir
make px4_sitl_spacecraft   # gz_atmos / gz_atmos_dual
make px4_sitl_uuv          # gz_uuv_bluerov2_heavy