ub_gazebo

This repository provides code for launching multiple robots in Gazebo (classic, v11). This was mostly just to document things for myself; if this info helps you, then I'll be thrilled.

---

Pre-reqs

This documentation was written for a system with these properties:

• ROS Noetic running on Ubuntu 20.04
• Gazebo Classic, Version 11
• The user's ROS catkin workspace is ~/catkin_ws/; this ub_gazebo package will be installed in ~/catkin_ws/src/ub_gazebo.

Currently, there is support for 2 kinds of robots: The turtlebot3 and Clearpath Robotics Husky.

Install turtlebot3 packages

• Start with the main packages:

  sudo apt-get update
  sudo apt --yes install ros-noetic-turtlebot3-msgs
  sudo apt --yes install ros-noetic-turtlebot3
  

• Now install the simulation package (see https://emanual.robotis.com/docs/en/platform/turtlebot3/simulation/):

  cd ~/catkin_ws/src/
  git clone -b noetic-devel https://github.com/ROBOTIS-GIT/turtlebot3_simulations.git
  cd ~/catkin_ws && catkin_make
  

• For convenience, save environment variables in ~/.bashrc. Use any text editor to open the file (e.g., pico ~/.bashrc). Then, add the following line to ~/.bashrc:

export TURTLEBOT3_MODEL=waffle_pi  # options are burger, waffle, waffle_pi

Install Husky packages

• Start with the main packages. See https://www.clearpathrobotics.com/assets/guides/noetic/ros/Drive%20a%20Husky.html

  sudo apt-get --yes install ros-noetic-husky-desktop
  sudo apt-get --yes install ros-noetic-husky-simulator 
  sudo apt-get --yes install ros-noetic-husky-navigation
  

• For convenience, save environment variables in ~/.bashrc. Use any text editor to open the file (e.g., pico ~/.bashrc). Then, add the following lines to ~/.bashrc:

  # Enable front/rear scanners on Husky:
  export HUSKY_LMS1XX_ENABLED=1
  export HUSKY_LMS1XX_SECONDARY_ENABLED=1
  
  # Enable Intel RealSense camera, and shift its location:
  export HUSKY_REALSENSE_ENABLED=1
  export HUSKY_REALSENSE_XYZ="0.35 0 0.01"
  

---

Install ub_gazebo (this repo)

cd ~/catkin_ws/src
git clone https://github.com/optimatorlab/ub_gazebo.git

• It is also recommended to checkout a branch, using your GitHub username as the branch name:

  cd ~/catkin_ws/src/ub_gazebo
  git checkout -b YOUR_GITHUB_USERNAME_GOES_HERE
  

Now, compile the ub_gazebo package:

cd ~/catkin_ws
catkin_make

Finally, add the following to ~/.bashrc:

# Update model/resource paths to let Gazebo know where to find some of our materials
export GAZEBO_MODEL_PATH=$GAZEBO_MODEL_PATH:~/catkin_ws/src/ub_gazebo/models
export GAZEBO_MODEL_PATH=$GAZEBO_MODEL_PATH:~/.gazebo/models
export GAZEBO_RESOURCE_PATH=$GAZEBO_RESOURCE_PATH:~/catkin_ws/src/ub_gazebo/launch/media
export GAZEBO_RESOURCE_PATH=$GAZEBO_RESOURCE_PATH:~/catkin_ws/src/ub_gazebo/worlds

# Let Python know where our model_launcher module is saved:
export PYTHONPATH=$PYTHONPATH:~/catkin_ws/src/ub_gazebo/scripts

---

Usage/Examples

The code in this repo is primarily for launching multiple robots/models in a Gazebo world. Two different approaches are described below:

Option 1 - Use a single .launch file to spawn a world with one or more robots

See world_plus_huskies.launch

roscd ub_gazebo/launch
roslaunch ub_gazebo world_plus_huskies.launch

See world_plus_turtlebots.launch

roscd ub_gazebo/launch
roslaunch ub_gazebo world_plus_turtlebots.launch

• NOTE: For some reason, Gazebo is complaining when we try to spawn both Huskies and Turtlebots. (Specifically, the message is a variant of [ WARN] [1711136096.606182878, 56.509000000]: Could not obtain transform from base_link to husky/0_tf/base_link. Error was "base_link" passed to lookupTransform argument source_frame does not exist..).

Option 2 - Spawn a world with one file, spawn robots separately

This approach works well if you want to add robots over time (instead of all at once). This also works well for adding robots programmatically (e.g., via a Python script).

You'll need (at least) two terminal windows:

Terminal 1 - Use a .launch file to spawn a world, but with no robots.

roscd ub_gazebo/launch
roslaunch ub_gazebo world_only.launch

The world_only.launch file can be edited to spawn different worlds.

Terminals 2, 3, 4, ... - Start simulated robots on demand.

Here are two ways to do this:

• Using one terminal window per roslaunch command, spawn the robot of your choice:
  • roslaunch ub_gazebo husky_spawn.launch robot_namespace:=husky/3 x:=1.3 y:=2.1 z:=0.0 yaw:=1.57 webcam_pitch:=0.0
  • roslaunch ub_gazebo turtlebot_spawn.launch robot_namespace:=tb/0 x:=2.3 y:=2.1 z:=0.0 yaw:=1.57
  • Just make sure each robot has a unique robot_namespace.
• Alternatively, you can edit the robot_launcher.py script and run it:

  roscd ub_gazebo/scripts
  python3 robot_launcher.py
  

  This will launch several Huskies or Turtlebots at once.

---

Adding Objects/Models (not robots) to Gazebo

The ~/catkin_ws/src/ub_gazebo/scripts/model_launcher.py module can be used to spawn objects in Gazebo, including:

• boxes
• spheres
• aruco tags
• traffic cones

The following Python script demonstrates how to create one of each of the above types of model.
Before running this script, you must have a Gazebo world running (e.g., roslaunch ub_gazebo world_only.launch).

import rospy
import model_launcher
import numpy as np

rospy.init_node('myNode', anonymous=True)

# Initialize a dictionary for all of our models/objects
objects = {}

# -----------------
# 1. Add an aruco tag
args = { 'markerID': 120, 
         'marker_width': 0.75, # [m]
         'board_width': 1.0}   # [m] (like 1.0)

objects['aruco120'] = model_launcher.launcher(objectType = 'aruco', 
					 gazeboModelName = 'aruco120', 
					 namespace = 'myobjects', 
                     modelYawOffsetRad = 0, 
                     x = 1, y = 2, z = 3, 
                     rollRad = 0, pitchRad = np.deg2rad(90), yawRad = 0, 
                     args=args)

# -----------------
# 2. Add a box
args = { 'color_r': 1.0,
		 'color_g': 0.0,
		 'color_b': 0.2,
		 'dim_x': 1.0, 
		 'dim_y': 0.1, 
		 'dim_z': 1.0, 
		 'useGravity': True}

objects['box1'] = model_launcher.launcher(objectType = 'box', 
					 gazeboModelName = 'box_1',
					 namespace = 'myobjects', 
                     modelYawOffsetRad = 0,					  
                     x = 2, y = 3, z = 4, 
                     rollRad = 0, pitchRad = 0, yawRad = 0, 
                     args=args)

# -----------------
# 3. Add a traffic cone
args = { 'color': 'orange',   # ("orange", "red", "yellow", "green", "blue") 
         'scale': 0.75}        # [m] (like 0.75).  Our cone model is 1m^3, so scale and size are equiv.

objects['cone1'] = model_launcher.launcher(objectType = 'cone', 
					 gazeboModelName = 'cone',
					 namespace = 'myobjects', 
                     modelYawOffsetRad = 0,					  
                     x = 2, y = 1, z = 0, 
                     rollRad = 0, pitchRad = 0, yawRad = 0, 
                     args=args)
                     
# -----------------
# 4. Add a sphere
args = { 'color_r': 0.1, 
		 'color_g': 0.8,
		 'color_b': 0.0,
		 'radius': 1.0, 
		 'useGravity': False}
		 
objects['green_ball'] = model_launcher.launcher(objectType = 'sphere',  
					 gazeboModelName = 'sphere_a',
					 namespace = 'myobjects', 
                     modelYawOffsetRad = 0,					  
                     x = 1, y = 1, z = 4, 
                     rollRad = 0, pitchRad = 0, yawRad = 0, 
                     args=args)


# -----------------
# 5a. Add a "general" object (one that has a .dae model)
# See http://models.gazebosim.org/ for some examples.
# Also see `~/.gazebo/models` for models that were already downloaded to your machine.
args = {'scale': 5}

objects['barrier'] = model_launcher.launcher(objectType = 'jersey_barrier',  
					 gazeboModelName = 'jb1',
					 namespace = 'myobjects', 
                     modelYawOffsetRad = 0,					  
                     x = 1, y = 1, z = 4, 
                     rollRad = 0, pitchRad = 0, yawRad = 0, 
                     args=args, 
                     hasDAE = True)


# 5b. Add a "general" object that just has a `model.sdf` file (no meshes/*.dae)
args = {}
objects['beer'] = model_launcher.launcher(objectType = 'beer',  
					 gazeboModelName = 'beer_1',
					 namespace = 'myobjects', 
                     modelYawOffsetRad = 0,					  
                     x = 1, y = 1, z = 4, 
                     rollRad = 0, pitchRad = 0, yawRad = 0, 
                     args=args, 
                     hasDAE = False)


# -----------------
# 6. Delete each model

objects['aruco120'].delete()
objects['box1'].delete()
objects['cone1'].delete()
objects['green_ball'].delete()
objects['barrier'].delete()
objects['beer'].delete()

---

Coming Soon

The following improvements are planned:

• Improved documentation
• Support for drones
  • PX4
  • Clover