Drone inspection missions in simulation

This tutorial runs a photovoltaic panel inspection mission in the Gazebo simulation of the inspection testbed (TB2). It covers configuring the drone world, creating a mission with the built-in TUI, running it, and visualising the panels in RViz.

Requirements

• Aerostack2 and the TB2 project installed (see TB2: Inspection Testbed).

• tmuxinator installed (gem install tmuxinator or apt install tmuxinator).

• All commands run from the project root (TB2_Panel_Inspection_Simulation/).

Figure 1 Per-drone Aerostack2 node stack: Knowledge Base and Mission Monitor (collective awareness, pink) alongside the core motion behaviors and hardware interface.

Configure the world

The world YAML defines the GPS origin and the initial position of each drone in the simulation. The launcher reads drone namespaces directly from this file, so adding or removing a drone here is enough to change the swarm size.

The default single-drone world is config/world.yaml:

/**:
  platform:
    ros__parameters:
      gps_origin:
        latitude: 40.4405287
        longitude: -3.6898277
        altitude: 100.0

drone0:
  platform:
    ros__parameters:
      vehicle_initial_pose:
        x: -2.0
        y: -4.0
        z: 0.0

To add a second drone, append a new entry following the same pattern, then pass the file with -w:

drone1:
  platform:
    ros__parameters:
      vehicle_initial_pose:
        x: 2.0
        y: -4.0
        z: 0.0

Pre-built world files for 3, 4 and 5 drones are in config/ (e.g. config/world_swarm.yaml, config/world4drones_solar.yaml).

Create a mission with the TUI

The TUI is the recommended way to create mission files. Launch it from the project root:

python3 tui_experiments.py

The TUI opens on the Spec Generator screen. Use the Experiment Runner tab (Tab key) to execute batches of pre-built missions; this tutorial focuses on the generator.

Steps to create a mission:

1. Press Add [a] to open the spec editor form.

2. Fill in the basic parameters:

   • Run name — used as the filename prefix for the generated files.

   • Arena bounds — the [x_min, x_max] × [y_min, y_max] extent of the Gazebo world in metres.

   • Drones — count — number of drones; must match the entries in your world YAML.

   • Drone start X — the X coordinate where drones are placed in a line at startup.

3. Configure the inspection areas. Choose a layout (grid_areas for a regular grid, strip_areas for horizontal bands, or custom to draw polygons interactively). For custom, a matplotlib canvas opens:

   • Left-click — add a vertex to the current polygon.

   • Right-click or n — close the current polygon and start a new one.

   • u — undo the last vertex.

   • Delete / Backspace — discard the current in-progress polygon.

   • Close window — confirm and return to the form.

4. Set the coverage parameters under World:

   • street_spacing — distance between adjacent coverage lanes (metres).

   • wp_space — distance between waypoints along each lane (metres).

   • height — inspection flight height (metres).

   • speed — coverage flight speed (m/s).

   • orientation — sweep direction in degrees (0° = along X, 90° = along Y).

5. Press Generate [g] to write the mission files without launching, or Generate & Run [r] to generate and immediately start the stack and mission.

The generator writes:

• config/exp_config/<name>/<name>.yaml — world file with drone initial positions.

• missions/<name>/<name>_count<N>.yaml — mission file for N drones.

Launch the simulation

1. Start the Aerostack2 stack, passing the world file that matches your drone count:

   ./launch_as2.bash -w config/exp_config/<name>/<name>.yaml
   

2. In a second terminal, open the ground station (RViz + monitoring):

   ./launch_ground_station.bash -w config/exp_config/<name>/<name>.yaml
   

Run the mission

Send the generated mission file to the running stack:

python3 send_mission.py missions/<name>/<name>_count<N>.yaml -s

The -s flag enables simulation time. drone0 acts as auctioneer: it plans the coverage waypoints for all areas, runs an auction with the collective awareness structure, and each drone executes the waypoints assigned to it.

Add -v for verbose output showing waypoint assignments and auction results.

View panels in RViz

When send_mission.py starts, it publishes the inspection panel meshes and the ground plane to RViz automatically. Two MarkerArray topics become active:

• /solar_panels — one mesh marker per inspection panel.

• /ground_plane — the ground surface mesh.

These are displayed automatically if the ground station RViz config includes MarkerArray displays for those topics. To add them manually in RViz, click Add → By topic → /solar_panels → MarkerArray (and repeat for /ground_plane).

To publish the panel markers standalone — for example when replaying a rosbag without a live mission — edit the mission path in publish_static_markers.py and run:

python3 publish_static_markers.py

Stop the simulation

./stop.bash