Project Overview
This repository contains the software developed by the College of DuPage team for the NASA Lunabotics competition. It is built for ROS 2 Humble on Ubuntu 22.04 for x86-64 architecture.
System Components
Computer
- ASRock 4X4 BOX-8840U w/ 64GB RAM
Sensors
- RPLidar S3
- RPLidar S2L
- Intel RealSense D455 Depth Camera
- Intel RealSense D456 Depth Camera
Hardware
- REV Robotics NEO Vortex (x2)
- REV Robotics Spark Max (x2)
- REV Robotics Power Distribution Hub
- Turnigy 14.8V 8000mAh LiPo Battery
- Turnigy 14.8V 12000mAh LiPo Battery
- ODrive USB-CAN Adapter
Installation
Note: You will need to have already installed ROS 2 Humble before continuing with installation. The guide can be found here. Install both ros-humble-desktop and ros-dev-tools.
(Optional) 1. Append lines to .bashrc
.bashrc is a script that runs everytime a new terminal window is opened and has various configurations, environment variables, and commands for setup. There is a bug in the VSCode terminal that will cause a symbol lookup error, so you have to unset the path variable using unset GTK_path. If you haven't already added source /opt/ros/humble/setup.bash to your .bashrc file, it simply runs the setup script for ROS 2 Humble.
This will permanently append these two lines to your .bashrc file, so there is no need to run it again. If you want to edit the file manually, use nano ~/.bashrc or gedit ~/.bashrc if you prefer a text editor GUI instead.
2. Setup workspace and clone repository
If you have previously cloned this repository and do not see anything in the folders located in third_party_packages after running git pull, run git submodule update --init --recursive --remote inside the lunabot_ros folder to initialize the submodules.
3. Install dependencies
Run the installation script to install the required dependencies. chmod +x gives permission for the script to be executable.
(Optional) 4. Install Foxglove Studio
If you would prefer to use Foxglove Studio instead of RViz2 to visualize the robot, you can install it with:
5. Build the workspace
Building may take some time due to the external packages in third_party_packages. Various flags such as -DRTABMAP_SYNC_MULTI_RGBD=ON need to be set to enable extra features for RTAB-Map.
To avoid building the entire workspace all over again after the initial build if you make changes, use colcon build --packages-select name_of_package and choose the package that you made changes to for rebuilding. You can list multiple packages after the --packages-select flag. You only need to rebuild the workspace if you modify a file for a compiled language such as C++ or add new files, the flag --symlink-install will automatically reflect the changes in Python, URDF, Xacro, and YAML files.
6. (Optional) Set MAKEFLAG and Rebuild
If your computer keeps crashing while trying to build, colcon build may be trying to do too many things at once. Setting this flag to -j1 limits each package's internal make jobs to 1 thread. You can either increase or reduce both this and --parallel-workers, increasing will make it build faster but may put more stress on your computer, leading to freezing.
Next, rebuild using the same commands in step 5. Build the workspace.
Simulating the Robot
The launch files have various parameters that can be set, such as changing the robot model, autonomy level, and choosing between RViz2 and Foxglove Studio for visualization. If you are using the parameter viz_type:=foxglove, refer to the Foxglove guide for connecting in the app. You can import the same layout I used by choosing Import from file... under the LAYOUT menu and selecting foxglove_layout.json from this directory.
A detailed list of the launch parameters can be found here.
1. Navigate to workspace and source setup
Open separate terminal windows and source the workspace setup for each next step:
2. Launch visualization
3. Launch mapping and navigation
Gazebo
RViz2
Foxglove Studio
Running the Physical Robot
SSH into Robot Computer
SSH (Secure Shell) allows you to access another device over the network and run commands. In this context:
- The client is your personal computer (e.g., your laptop).
- The host is the robot's onboard computer (e.g., ASRock 4X4 BOX-8840U).
1. Install and enable SSH server (host):
2. Install SSH client (client):
3. Create SSH-key (client)
4. Get username and IP address (host)
This will return the username of the host, although you can also see the username just by looking at the terminal. It is the first name before the @, for example, the username would be asrock for asrock@asrock-main.
Next, get the IP address:
The IP address is the first set of numbers in the list.
5. Establish SSH connection (client)
Using the username and IP address from the previous step, now you can connect to the host. It may look something like this for example:
It will ask you if you are sure you want to connect, type yes. Then, confirm by typing in the host's password.
Configure Device Permissions
1. Add user to dialout group then restart (host)
Use ls /dev/ttyUSB* to identify device numbers if the lidars are disconnected and reconnected, then adjust the lidar "serial_port" parameters in real_launch.py accordingly.
2. Setup camera udev rules (host)
Make sure all cameras are unplugged while setting up the udev rules.
Running Launch Files
1. Source workspace setup (both client and host)
2. Connect controller and run joy node (client)
Connect your controller either through a wired or Bluetooth connection to the client.
