Skip to content

Ground Control Station (GCS)

The Ground Control Station provides monitoring, control, and mission planning capabilities for AirStack robots. Operators use the GCS to:

  • Monitor robot status, camera/depth feeds, and sensor streams
  • Send task commands (takeoff, land, navigate, fixed trajectory, search, exploration)
  • Draw waypoint routes and polygon areas directly on the map
  • Visualize robot poses, planned paths, and maps for the whole fleet
  • Record mission data as ROS 2 bags

The main operator interface is Foxglove Studio, extended with custom AirStack panels. Everything runs inside one Docker container.

Directory Structure

The GCS is organized under gcs/:

gcs/
├── docker/                           # GCS containerization
│   ├── docker-compose.yaml           # gcs (sim/dev) + gcs-real (field) services
│   ├── gcs-base-docker-compose.yaml  # Shared base service (command, env, mounts)
│   ├── Dockerfile.gcs                # Image: ROS 2 Jazzy + Foxglove Studio + DDS Router
│   ├── .bashrc                       # Shell config mounted into the container
│   └── Foxglove/                     # Foxglove Studio app state (mounted, gitignored)
├── foxglove_extensions/              # Custom Foxglove panels + layout tooling
│   ├── waypoint-editor/              # Click-to-place waypoint routes
│   ├── polygon-editor/               # Click-to-draw polygon areas
│   ├── robot-commands/               # "Robot Tasks" command panel
│   ├── install.py                    # Installs the panels into Foxglove on startup
│   ├── render_layout.py              # Renders the NUM_ROBOTS-matched layout
│   └── airstack_default.json         # Single-robot layout template
├── ros_ws/                           # ROS 2 workspace
│   └── src/
│       ├── action_relay/             # Bridges task actions from GCS domain 0 to each robot domain
│       ├── gcs_visualizer/           # Fleet markers/poses for Foxglove's 3D panel
│       └── common/                   # Mount of common/ros_packages (desktop_bringup, coordination, ...)
├── saves/                            # Persisted waypoint/polygon editor saves (mounted at /root/.airstack)
└── bags/                             # Recorded mission data (mounted at /bags)

Launch Structure

The GCS is launched via Docker Compose (gcs/docker/docker-compose.yaml, extending gcs-base-docker-compose.yaml). On container start it:

  1. Restarts the SSH daemon
  2. Installs the custom Foxglove extensions (foxglove_extensions/install.py)
  3. Renders the multi-robot Foxglove layout to /root/airstack_layout_num_robots_<N>.json (render_layout.py)
  4. Opens a tmux session named bringup
  5. If AUTOLAUNCH=true, runs ros2 launch desktop_bringup gcs.launch.xml in that session

desktop_bringup/launch/gcs.launch.xml (in common/ros_packages) starts:

Component Purpose
Foxglove Studio (desktop app) Main operator GUI, connects to ws://localhost:8765
foxglove_bridge ROS 2 ⇄ Foxglove WebSocket bridge on port 8765
Gossip bridge Connects GCS domain 0 to the gossip bus (domain 99)
gcs_visualizer Renders per-robot meshes/trajectories/maps in a shared global ENU frame
action_relay One relay per robot: forwards task goals from Foxglove to each robot's domain

Launch command:

# Start GCS container (usually alongside the rest of the stack)
airstack up gcs

# Attach to the container's tmux session
airstack connect gcs

Learn more: Docker Configuration

GCS Interfaces

Foxglove Studio

The operator GUI. The container runs the Foxglove Studio desktop app on the host's X display; you can also open Foxglove on the host and connect to the bridged WebSocket.

Features:

  • 3D scene visualization: fleet poses, trajectories, global plans, VDB maps (details)
  • Robot Tasks panel: send takeoff / land / navigate / trajectory / search / exploration commands per robot
  • Waypoint & polygon editors: click-to-place routes and areas (guide)
  • Per-robot tabs: the rendered layout creates one tab per robot for NUM_ROBOTS robots
  • Data recording and playback

Connection: ws://localhost:8765 inside the container, ws://localhost:8766 from the host (the gcs service publishes 8766→8765). See GCS Foxglove Visualization for the layout import flow.

System Requirements

Hardware:

  • Hard Disk: 60GB free space
  • RAM: 8GB minimum, 16GB recommended
  • CPU: 4 cores minimum, 8+ recommended
  • GPU: NVIDIA GPU (the service reserves one for rendering)
  • Network: access to robot containers (via airstack_network or the LAN for gcs-real)

Software:

  • OS: Ubuntu 22.04/24.04 LTS
  • Docker: installed via airstack install
  • Display: X11 display server (for the Foxglove Studio window)

Quick Start

  1. Launch the stack (sim + robots + GCS — the desktop profile includes the gcs service): 

    airstack up --sim isaac --robots 1

  2. Foxglove opens automatically in the GCS container (when AUTOLAUNCH=true). Import the rendered layout and connect — see GCS Foxglove Visualization.

  3. Command robots from the Robot Tasks panel; place waypoints with the editors.

Full tutorial: Getting Started

Next Steps