Mahmoud Abas, Full-Stack Software Engineer & AI

ROS 2 Practical - TurtleBot3 Lab 07 - Autonomous Navigation with Nav2

In the print dialog, choose "Save as PDF" as the destination.

TurtleBot3 - Lab 07

Autonomous Navigation with Nav2

From a saved map to autonomous goal-reaching.

Learning Objectives

Understand the `Nav2` navigation stack architecture
Learn about `AMCL` (Adaptive Monte Carlo Localization)
Experience autonomous path planning and execution
Practice setting navigation goals using RViz
Builds directly on Lab 06's `SLAM` map

Setup and Launch

bash

sudo apt install ros-jazzy-navigation2 ros-jazzy-nav2-bringup

colcon build --packages-select turtlebot3_lab07
source install/setup.bash

export TURTLEBOT3_MODEL=waffle_pi
ros2 launch turtlebot3_lab07 lab07.launch.py

One command launches Gazebo, the full Nav2 stack, and RViz
Use your own Lab 06 map: map:=$HOME/maps/my_map.yaml

Nav2 Components

Component	Description
`AMCL`	Estimates robot pose on a known map
Map Server	Loads and serves the occupancy grid
Planner Server	Computes global paths start to goal
Controller Server	Executes path with local obstacle avoidance
`BT Navigator`	Behavior-tree-based navigation orchestration

Nav2 Architecture Flow

RViz Goal

2D Goal Pose

BT Navigator

Planner

Global

Controller

Local

Behavior

Recovery

Global Costmap

Local Costmap

AMCL

Localize

The BT Navigator orchestrates planning, control, and recovery behaviors as a behavior tree

Required

Step 1: Set the Initial Pose

The robot doesn't know where it is on the map until told

bash

ros2 topic pub /initialpose geometry_msgs/msg/PoseWithCovarianceStamped \
  "{header: {frame_id: 'map'}, pose: {pose: {position: {x: -2.0, y: -0.5, z: 0.0}, orientation: {w: 1.0}}}}" --once

Click 2D Pose Estimate in RViz
Click the robot's real location on the map
Drag to set the facing direction
Watch the particle cloud converge

Step 2: Send a Navigation Goal

Click 2D Goal Pose in RViz
Click the destination on the map
Drag to set the final orientation

The robot plans a global path and drives there autonomously, avoiding obstacles locally

bash

ros2 action send_goal /navigate_to_pose nav2_msgs/action/NavigateToPose \
  "{pose: {header: {frame_id: 'map'}, pose: {position: {x: 0.5, y: 0.5}, orientation: {w: 1.0}}}}"

Command-line equivalent

Key Topics

Topic	Purpose
`/map`	Static map from the map server
`/scan`	LiDAR data feeding `AMCL` localization
`/cmd_vel`	Velocity commands to the robot
`/goal_pose`	Navigation goal from `RViz`
`/plan`	Computed global path

Troubleshooting

Frozen or static Gazebo or RViz: Confirm the workspace is sourced, wait about 30s for init, check /clock is advancing

Robot not localizing: Verify the initial pose was set correctly; drive slowly to help AMCL's particle filter converge

Navigation fails to plan: The goal must be in free (white) space; try a closer goal first

Waiting for Transform: Wait longer for startup; confirm all Nav2 nodes are running

bash

ros2 node list | grep -E "amcl|planner|controller|bt_navigator"
ros2 run tf2_tools view_frames

Key Takeaways

Nav2 = AMCL (localization) + Planner + Controller + BT Navigator, all orchestrated as a behavior tree
A saved SLAM map (Lab 06) is Nav2's input, navigation needs a map before it can plan
Setting the initial pose is mandatory; skipping it is the number one reason navigation fails
Global costmap plans the overall path; local costmap handles real-time obstacle avoidance
RViz's 2D Pose Estimate and 2D Goal Pose tools are the primary way to interact with Nav2 in the lab
Diagnosing Nav2 issues starts with ros2 node list (are AMCL, planner, controller, BT Navigator running?) and the TF tree

Open this lesson

Mahmoud AbasTurtleBot3 Lab 07 - Autonomous Navigation with Nav2

sudo apt install ros-jazzy-navigation2 ros-jazzy-nav2-bringup colcon build --packages-select turtlebot3_lab07 source install/setup.bash export TURTLEBOT3_MODEL=waffle_pi ros2 launch turtlebot3_lab07 lab07.launch.py

Component

Description

AMCL

Estimates robot pose on a known map

Map Server

Loads and serves the occupancy grid

Planner Server

Computes global paths start to goal

Controller Server

Executes path with local obstacle avoidance

BT Navigator

Behavior-tree-based navigation orchestration

Topic

Purpose

/map

Static map from the map server

/scan

LiDAR data feeding AMCL localization

/cmd_vel

Velocity commands to the robot

/goal_pose

Navigation goal from RViz

/plan

Computed global path