Skip to content

ROS 2 Software Stack for Autonomous Navigation

This core software lesson guides students through building the "brain" of an autonomous mobile robot (AMR) using the ROS 2 framework. Students implement key autonomy modules such as localization, mapping, and path planning using open-source ROS 2 packages. The final result is a fully integrated navigation system capable of moving through unknown environments safely and intelligently.

ROS 2 UI architecture Overview

Core Concepts

  • Configuring and tuning the Navigation2 (Nav2) stack
  • Implementing odometry, TF tree broadcasting, and localization with AMCL
  • Performing mapping and SLAM using tools like slam_toolbox or cartographer
  • Setting up sensor drivers (e.g., LIDAR, IMU, encoders) and robot model (URDF/Xacro)
  • Developing and launching ROS 2 nodes in Python or C++
  • Performing simulation testing in Gazebo and RViz

System Overview

  1. The robot’s physical model is described using URDF or Xacro, including joints, wheels, and sensors.
  2. Sensor drivers are launched to stream real-time data from LIDARs, IMUs, and wheel encoders.
  3. Odometry and TF frames are published, forming the basis of robot localization.
  4. The AMCL node is used for localization against a map, or SLAM is used for building maps dynamically.
  5. The Nav2 stack plans paths, avoids obstacles, and sends velocity commands to the base controller.
  6. Simulation and visualization are conducted using Gazebo and RViz, with testing routines and tuning cycles.

Learning Outcomes

  • Set up and configure a modular ROS 2 software stack
  • Develop and launch ROS 2 nodes in both Python and C++
  • Build real-time localization and mapping pipelines
  • Understand and manipulate the TF2 tree and ROS message types
  • Integrate and test robot software in simulated environments
  • Gain practical skills in the ROS 2 development workflow

Optional Extensions

  • Replace AMCL with EKF-based localization using robot_localization package
  • Use Nav2 Behavior Tree customization to define custom navigation logic
  • Add multi-robot simulation scenarios in Gazebo
  • Deploy and benchmark on a physical robot for real-world tuning
  • Create a teleoperation or diagnostic GUI for remote mission support