Getting Started

At a high level, to install and run this projects code on your computer or VM involves:

1. Pulling the 3 main packages (arm_master, brick_manager, spawn_manager) from github and building them
2. Pulling relevant packages for interfacing with Panda robot (franka_ros, franka_gazebo)
3. Downloading relevant support files (brick.stl)

We will now walk step by step to go from an empty workspace to a robot that can build a wall in Gazebo

Set Up

Before installing the project’s packages we need to make sure everyone has set up their environment set up correctly and are using the correct tools. We assume you have a basic knowledge of ROS.

Note

The project was developed using ROS Kinetic running on a native Ubuntu 16.04 os. It was also tested on a Ubuntu 18.04 VM running ROS Melodic. Given ROS’s backwards comparability it is likely that it should work on your setup as well, but having a similar setup is likely to minimize possible problems.

First you will need to install catkin tools. Catkin tools is the preferred build system used by the ASL lab at ETHz. It provides additional functionality (catkin clean) and greater control over build configs.

Please refer to documentation for details on installation and usage

Warning

The default catkin_make and catkin tools build systems are not compatible. It is not recommended to mix the two. Please make sure you only use the catkin build command to build your src.

Note

If for some reason you end up mixing the two or the project generally begins to break down, you can manually reset it by deleting the build, devel and logs folders. Running the catkin clean command from terminal provides similar functionality.

We will also add some code to your .bashrc, to automate some of the background installation work. From your home directory, open .bashrc with your favorite text editor:

gedit ~/.bashrc

Copy paste the following two lines at the bottom:

source /opt/ros/kinetic/setup.bash
source ~/de3_ws/devel/setup.bash

Then close and save.

Assuming you now have catkin tools installed and your .bashrc edited, we can begin the setup process. First create a new workspace and initialize it. From within your desired directory:

mkdir de3_workspace
cd de3_workspace
mkdir src
catkin init

Build the workspace:

catkin build

You should now see in addition to your src folder, build, devel and logs folders.

Warning

Remember to source your workspace after building to let ROS know you have made changes. From your workspace directory:

source devel/setup.bash

We list it here once and assume you will do it after each catkin build in this tutorial. Note that generally during development you will not need to re-source your workspace, especially when writing python scripts. You are required, however to re-source your workspace when creating new packages or messages.

As a rule of thumb, if ROS commands like roscd aren’t working, and you are getting module not found errors when you try to import custom messages, try and also re-source your workspace.

Note

In the setup we added the following code to your .bashrc

source /opt/ros/kinetic/setup.bash
source ~/de3_ws/devel/setup.bash

The .bashrc is run every time you start a new terminal session. That means instead of manually re-sourcing your workspace, you can simply close and open a new terminal, and ROS will update it self. If you decide to continue working in the same terminal session, however, you will need to run the aforementioned code.

Congrats, you have created and built your ROS workspace! In the src folder is where you now will install all relevant ROS packages.

Installing Project Packages

First we install the custom packages written for this project. Find links to the three packages and download using git

• arm_master
• brick_manager
• spawn_manager
• de_msgs

Example for arm master. First move into src directory

cd de3_workspace/src

Then clone

git clone https://github.com/de3-robo/arm_master.git

Note

Use ctrl-shift-v to paste the github link into terminal

At this point you should now have all 4 packages installed in your workspace/src folder. Lets build what we have to make sure its working

catkin build

If it worked, great job!

Installing 3rd Party Packages

Now we will install the 3rd party packages required for path planning and interfacing with Panda. These may or may not already be installed in your computer.

Note

Here we will only install necessary packages required to run Panda in Gazebo. Refer to the Running on Real Panda section to see setup for running on the real Panda.

First we download MoveIt. Please refer to MoveIt documentation for installation guide. We can also highly recommend walking through the MoveIt python tutorial.

You will now need to install:

• franka_gazebo
• panda_moveit_config

as before use the git command line calls and then build the package:

git clone https://github.com/de3-robo/franka_gazebo.git
...

catkin build

Great job! your almost there.

Installing Additional Files

First we will install the brick which is used to place build the wall in Gazebo.

Note

Note the brick provided on the course box folder needs to have its friction and mass parameters tuned to actually work in simulation. The brick we provided is not perfect, but should be sufficient as a starting point.

Navigate to your models folder in your .gazebo installation. On my computer .gazebo is accessible from the home directory.

cd .gazebo/models/

Then clone the brick into the brick model directory:

git clone https://github.com/de3-robo/Brick.git

Finally we need to install a new gazebo world and launch file with tables like those found on level 3 in the Dyson School. Download the code from the following table repo into your computer:

• Move file wall_table_world.launch into /opt/ros/kinetic/share/gazebo_ros/launch directory. Make sure to enter in your correct ROS version.
• Move file wall_table.world into /usr/share/gazebo-9/worlds directory.

If you made it this far, you’re doing great! Lets run the code now.

Running the Code in Gazebo simulation

Warning

This assumes that you already have Gazebo installed within Ubuntu / virtual machine.

We will now go through how to launch the actual simulation in Gazebo, which is a virtual simulation with laws of physics applied to objects within the environment, such as the Panda arm and bricks. This allows for the code to be safely tested in an environment with no risk of physical damage, before it is actually run on a real robot.

All the code can be run by 3 launch files located in the arm_master package. Make sure to run the code in order listed. In 3 separate terminals copy-paste the following commands:

roslaunch arm_master panda_one_brick.launch

Warning

This part is essential. The code will not work if you do not have the MoveIt motion planner module activated in rviz. To activate, click on add at the bottom of the display and then add the motion planning module. Use crtl-s to save the rviz config. Refer to installation video for more details.

Warning

Wait for each launch file to fully load before launching the next one.

roslaunch arm_master sim.launch

roslaunch arm_master light.launch

Additional Resources

Great intro to ROS by ETHz Autonomous Systems Lab

API reference for python movegroup interface, (has all the useful command you need for using MoveIt with python):