Simulation

Overview

Simulation allows you to test Pixel Robot’s software stack in a virtual environment before running it on real hardware. Pixel Robot uses Ignition Gazebo (Gazebo Fortress) integrated with ROS 2 through ros_gz_sim to provide a realistic simulation of sensors, robot motion, mapping, and navigation.

Simulation is strongly recommended for: - Testing navigation parameters - Debugging SLAM and localization - Developing new features safely - Learning ROS 2 workflows without hardware risk

Simulation Stack

The simulation setup consists of:

  • Ignition Gazebo (Gazebo Fortress)
  • ros_gz_sim for ROS 2 ↔ Gazebo integration
  • Simulated Pixel Robot model
  • Simulated sensors (LIDAR)
  • RViz2 for visualization

Setup

Installing Ignition Gazebo (Gazebo Fortress)

Install Ignition Gazebo Fortress:

sudo apt update  
sudo apt install ignition-fortress

Verify installation:

ign gazebo --version

Installing ROS–Gazebo Integration

Install the ROS 2 Gazebo integration packages:

sudo apt install ros-humble-ros-gz-sim  
sudo apt install ros-humble-ros-gz-bridge  
sudo apt install ros-humble-ros-gz-image

These packages enable communication between ROS 2 and Ignition Gazebo.

Preparing the Workspace

Ensure your Pixel Robot workspace is built and sourced:

cd ~/titan_ws  
colcon build  
source install/setup.bash

NOTE: If running simulation in a virtual machine:

export LIBGL_ALWAYS_SOFTWARE=1
export IGN_RENDER_ENGINE=ogre
export GZ_SIM_RENDER_ENGINE=ogre

This is used to make CPU use ogre renderer to launch gazebo.

Launching the Simulation

Start the simulation using the Pixel Robot simulation bringup:

ros2 launch pixel_simulation bringup_sim.launch.py

This launch file performs the following actions:

Robot Gazebo

  • Starts Ignition Gazebo
  • Loads the warehouse world
  • Spawns the Pixel Robot model
  • Launches simulated LIDAR
  • Bridges Gazebo sensor data to ROS 2
  • Starts RViz2 with a preconfigured view

Verifying the Simulation

Robot Rviz

Once launched, verify the following:

  • The robot appears in the Gazebo warehouse world
  • RViz2 shows the robot model
  • LIDAR scan data is visible in RViz2
  • TF frames are correctly published

At this stage, the robot is ready for mapping.

Mapping in Simulation

Launching SLAM

Open a new terminal, source the workspace, and launch SLAM:

source ~/titan_ws/install/setup.bash  
ros2 launch pixel_simulation slam.launch.py

This launch file runs SLAM Toolbox in online asynchronous mode, using simulated LIDAR data.

Visualizing the Map

Robot Mapping

In RViz2:

  • Set Fixed Frame to map
  • Add the Map display if not already visible
  • Observe the map being generated as the robot moves

You can use teleoperation to explore the environment and build the map.

Saving the Map

Once mapping is complete:

  1. In RViz2, open the Slam Toolbox Plugin
  2. Click Save Map
  3. Choose a directory and map name

This generates a .yaml and .pgm map file that can be reused for navigation.

After saving the map, stop the SLAM launch and start navigation.

Run the following command in a new terminal:

ros2 launch pixel_simulation pixel_nav.launch.py use_sim_time:=true map:=/absolute/path/to/your/map.yaml

This launch file starts:

  • Map server
  • Localization
  • Navigation stack (Nav2)
  • Costmaps and planners

The use_sim_time argument ensures synchronization with simulation time.

Sending Navigation Goals

Robot Navigation

In RViz2:

  1. Set Fixed Frame to map and give 2D Pose Estimate
  2. Use the 2D Goal Pose tool
  3. Select a goal location in the map

The robot will autonomously navigate to the goal within the simulated environment.

Simulation Time

Simulation runs on a simulated clock.

Ensure that: - use_sim_time is set to true - Nodes are synchronized to Gazebo time

This is critical for correct SLAM and navigation behavior.

Common Simulation Issues

Issue Possible Cause Solution
Robot not moving Controllers not loaded Check bringup launch logs
No LIDAR data Bridge not running Verify ros_gz bridges
Navigation fails Map not loaded Check map file path
Time-related errors use_sim_time false Enable simulation time

Simulation vs Real Robot

Key differences to keep in mind:

  • Simulation has perfect sensor alignment
  • No wheel slip unless modeled
  • Odometry is ideal compared to real robot
  • Navigation may appear smoother in simulation