Skip to main content
The TurtleBot3 development container includes three pre-configured simulation worlds, each designed for different testing scenarios and skill levels.

​
Available worlds

​
Empty world

A minimal simulation environment with a flat ground plane and no obstacles. 
tb3_empty
Use cases:
  • Testing basic robot movement
  • Learning teleoperation controls
  • Verifying sensor functionality
  • Debugging without environmental complexity
  • Quick simulation startup for development
Characteristics:
  • Fastest loading time
  • Minimal computational resources
  • Clean environment for isolated testing
  • Ideal for beginners
Start with the empty world when learning the basics or testing new code before moving to more complex environments.

​
TurtleBot3 world

An environment specifically designed for TurtleBot3 featuring obstacles, barriers, and confined spaces. 
tb3_world
Use cases:
  • Testing obstacle detection and avoidance
  • Navigation algorithm development
  • SLAM in structured environments
  • Collision detection testing
  • Path planning validation
Characteristics:
  • Multiple obstacles of varying sizes
  • Narrow passages for precision navigation
  • Moderate computational requirements
  • Balanced complexity for intermediate users
This world is excellent for testing navigation algorithms before deploying to more complex environments.

​
House environment

A detailed indoor house simulation with multiple rooms, furniture, and realistic indoor layout. 
tb3_house
Use cases:
  • Advanced SLAM mapping
  • Complex autonomous navigation
  • Multi-room exploration
  • Realistic indoor robotics scenarios
  • Final testing before real-world deployment
Characteristics:
  • Multiple interconnected rooms
  • Furniture and household objects
  • Realistic indoor environment
  • Higher computational requirements
  • Suitable for advanced testing
The house environment requires more system resources. Ensure your Docker container has sufficient CPU and memory allocated (4+ cores, 8GB+ RAM recommended).

​
Launching simulation worlds

​
Using aliases

The quickest way to launch any world: 
tb3_empty

​
Using full commands

If you prefer explicit commands or need to customize launch parameters: 
ros2 launch turtlebot3_gazebo empty_world.launch.py

​
Accessing the simulation

After launching any world, access the Gazebo GUI through the VNC desktop.
1

Launch a world

Start your chosen simulation environment:
tb3_world
2

Open VNC in browser

Navigate to the VNC desktop in your web browser:
http://localhost:6080
3

Enter password

When prompted, enter the VNC password:
ros
4

View Gazebo

You should see the Gazebo simulator running with your chosen world and the TurtleBot3 robot.

​
Switching between worlds

To switch to a different world, you must stop the current simulation first.
1

Stop current simulation

Press Ctrl+C in the terminal where the simulation is running.
2

Launch new world

Start the desired world:
tb3_house
If the new simulation doesn’t start cleanly, kill any remaining Gazebo processes:
pkill -9 gz

​
Software rendering mode

If you experience crashes or poor performance with GPU rendering, use software rendering variants. 
tb3_empty_sw
These aliases automatically set LIBGL_ALWAYS_SOFTWARE=1 for stable software-based rendering.

​
Choosing the right world

​
For beginners

Start with the empty world to:
  • Learn keyboard teleoperation
  • Understand ROS2 topics and nodes
  • Test basic robot movement
  • Familiarize yourself with the environment

​
For intermediate users

Use TurtleBot3 world to:
  • Practice SLAM mapping
  • Test navigation algorithms
  • Learn obstacle avoidance
  • Develop path planning skills

​
For advanced users

Use house environment to:
  • Create complex maps
  • Test multi-room navigation
  • Simulate realistic scenarios
  • Validate production-ready algorithms

​
Troubleshooting

​
Simulation won’t start

1

Verify TURTLEBOT3_MODEL

Check that the robot model is set:
echo $TURTLEBOT3_MODEL
Should output: burger
2

Kill existing processes

pkill -9 gz
pkill -9 gzserver
pkill -9 gzclient
3

Relaunch simulation

tb3_empty

​
No robot visible in Gazebo

1

Source workspace

sb
2

Verify packages

ros2 pkg list | grep turtlebot3
3

Rebuild if needed

cb
sb

​
Poor performance

  • Allocate more resources in Docker Desktop (Settings → Resources)
  • Use software rendering mode with _sw aliases
  • Choose a simpler world (empty instead of house)
  • Close unnecessary applications on your host system

Build docs developers (and LLMs) love

Get started for freeTalk to us