Context

As part of a research internship project, a modular ROS architecture was designed to enable cooperative navigation between heterogeneous autonomous vehicles: drones (UAVs) and ground robots (UGVs).
The contribution focused on supporting the development and implementation of this system, which aimed to be scalable, with dynamic task allocation and coordinated execution across different hardware platforms.

The architecture uses a unified communication protocol, distributed ROS nodes, and a namespace layout allowing multiple robots to be managed from a central ROS master.

Video 1 - Video demo: multi-robot cooperation.

In the following Video 1: Video demo: multi-robot cooperation. , the drone scans the area above the robots. It detects the ArUco markers designating the target positions, and the robots then move to these positions.

Core Architecture

Overview

The system is organized around three main components:

  1. Central ROS Master (PC)
    Hosts coordination logic and global launchers. Acts as scheduler and monitor.

  2. UGVs (ROS-based robots)
    Each UGV runs both common and robot-specific ROS nodes for navigation, teleoperation, and TCP communication.

  3. Drones (TCP clients)
    Lightweight platforms communicating with UGVs via TCP. Each drone exchanges structured task data and state updates with a specific ground robot.

Robot Architecture

The ROS robot stack is split into two layers:

  • common modules shared by all UGVs
  • specific modules tied to the hardware and control architecture of each UGV
Fig. 1 - Robot ROS architecture.

The Fig. 1 shows the ROS nodes and interfaces used in the project.

Robot-Specific Packages

These are located in the robots/ directory and provide hardware interfaces and configurations for each robot type (e.g., Turtlebot, Jetracer).

Communication Model

Drones and UGVs exchange data over TCP through a dual-channel setup:

  • Drone → UGV
    The drone acts as a client and sends mission data such as target coordinates and task instructions.

  • UGV → Drone
    The UGV hosts a TCP server to publish status and connection details (IP and port) back to the drone.

This bidirectional link supports dynamic tasking, status reporting, and reliable coordination between airborne and ground agents.

Trame Management

Data packets (“trames”) follow a consistent management process across robots:

  • Trame Content
    Defines the structure for tasks, targets, and status messages.

  • Trame Distribution
    Sends messages to individual robots, ensuring isolated task execution in multi-robot scenarios.

  • Trame Sequence
    Preserves ordering to ensure predictable and safe execution.

Fig. 2 - Trame (data packet) sequence.

The Fig. 2 illustrates the packet flow and ordering in task execution.

Namespace and Multi-Robot Configuration

Each robot operates under a dedicated namespace (for example, /turtlebot/robot) to keep topics and services isolated. This setup is essential for running multiple identical robots in parallel.

ROS networking follows the Multiple Machines configuration, with the PC as master and robots as remote clients. Each device must be configured with the appropriate ROS_MASTER_URI and ROS_IP values for the network.