Team Inspiration's repository for the RoboSub 2023 competition. We are using ROS, python, and C++.
Please use this repository only for source files; do not upload binary files like images/videos/ROSBags.
We have two different subs, Onyx and Graey, they both run on similar software setups. (compat issues with Jetson Nano and Jetson NX have been mostly mitigated)
- Ubuntu 20.04
- ROS Noetic
- Python 3.8.10
- OpenCV 4.7.0 (CUDA)
- CVbridge (Noetic)
- Ubuntu 18.04
- Ros Melodic
- Python 3.8.0 (non-native)
- OpenCV 4.7.0 (CUDA)
- CVbridge (Noetic modified)
Because of this, when we design our code, we need to make sure that it is compatible between both subs.
Note: If using installScripts the first two steps will be done for you, you just need to install
(for the following, you can also use the https link instead of the ssh link)
git clone [email protected]:InspirationRobotics/robosub_2023.git
mv robosub_2023 auv && cd auv
python3 -m pip install -e .if operating Onyx, you will need this additionnal step to install the dependancy for the DVL:
git clone https://github.com/waterlinked/dvl-python.git
cd serial
cd dvl-python
pip install -e .
#### On your own computer
```bash
git clone [email protected]:InspirationRobotics/robosub_2023.git
cd robosub_2023
python3 -m pip install -e ."[cv,dev]"
# or
python2 -m pip install -e ."[cv,dev]"In order not to have a messy repo, we have a specific structure for the repo. Please follow this structure when adding new features.
robosub_2023
├── auv
│ ├── device
│ │ ├── # everything related to the sensors
│ │ ├── # each device should have its own file/folder
│ ├── localization
│ │ ├── # TODO
│ ├── mission
│ │ ├── # Mission classes
│ │ ├── # see auv/mission/template_mission.py for an example
│ ├── motion
│ │ ├── # Actuators code
│ │ ├── # each actuator should have its own file/folder
│ ├── cv
│ │ ├── # All CV classes and functions
│ │ ├── # see auv/cv/template_cv.py for an example
│ ├── utils
│ │ ├── # utility functions
├── scripts
│ ├── # bash scripts
├── mission
│ ├── # Mission files (calling mission classes)
├── tests
│ ├── # Unit tests (pytest)
├── config.json # where all the config values used in mission files are stored
├── setup.py # setup file for pip install
├── .gitignore
├── README.mdThis will run all the tests in the tests folder using pytest.
Any file that has the name test_*.py will be run.
pytestWe have a mission class for each individual mission/task.
Those classes are then called in a mission file (see mission/ folder).
The mission file will be the one giving the mission class the necessary parameters to run the mission (for examples variables from the config file).
You can run individually mission classes for debugging purpose (if they have a if __name__ == "__main__" block) by running:
python3 -m auv.mission.<mission_name>
# or
python2 -m auv.mission.<mission_name>Else, you can import the mission class in a mission file and run it from there.
from auv.mission.<mission_name> import <MissionClass>
# initialize mission
mission = <MissionClass>()
# run mission
mission.run()
# cleanup mission after it is done
mission.cleanup()There exists a mission planner that allows you to load and run missions from a yaml file. This is useful to craft a sequence of missions to run in a competition while avoiding the need to modify the code. It is a graph based planner that allows you to define the sequence of missions and the conditions to run them. On top of that, it does catch exceptions (and thus not leave the sub in an unknown state).
You can also visualize with the option -v the graph and double check that the missions are in the right order.
To run the mission planner, you can use the following command:
python3 -m auv.mission.mission_planner -p <path_to_yaml_file> (-v optional for graph visualization)You can find an example of a yaml file in the missions/ folder here: plan_template.yaml.
To fully understand how to use the mission planner, you can take a look at the mission_planner.py file.
The idea behind the CV classes is that they are run in a separate thread and they publish their results to a ROS topic.
You can run individually CV classes for debugging purpose (if they have a if __name__ == "__main__" block) by running:
python3 -m auv.cv.<cv_name>In order to contribute to the repo, you will need to open pull requests. To do so, create a new branch with ideally this format: <your_name>/<feature_name>.
Then, when you are done with your feature, open a pull request on github.
Opening a pull request will trigger a code review from one of the team members. This team member will be in charge of reviewing your code and making sure that it is working as intended. Please provide as much information as possible in your pull request (what you changed, why you changed it, etc...). Keep up the good practices !
What if the default branch has changed since I started working on my feature ? No problem, just rebase your branch on the default branch and push it again. This will update your branch with the latest changes from the default branch.
git checkout <your_branch> git rebase origin/master git push origin <your_branch>
If it's a small change / fix or in a period of rush, of course you can push directly to master, but please try to avoid it as much as possible, we don't want to push something that is not working into the sub especially right before testings.
soon, we will have a simulator that will act as a digital clone for the sub, you will be able to spinup a docker container and test your code in it.
Additionally, you can use pytest for unit tests. Please write unit tests for your code and make sure that they pass before pushing your code. This is to make sure that we don't break anything when we add new features. Tests are mainly there to make sure that there is no syntax error and that the code runs without crashing, not necessarily to make sure that the code is doing what it is supposed to do (although it is a plus).
It is always pleasant to read code that is well formatted and well commented. Ideally we would want to format our code using black, so try to run regularly on your code:
black .If you have trouble running this command, you can use the following command instead:
python3 -m black .(bonus) Code analysis is done using pylint. we will omit C0115 and C0116 for now, but try to fix the other errors.
pylint --disable=C0115,C0116 auv-
Git pull if necessary
cd auv git pull -
First, start running roscore and mavros. This script will launch both at the same time in separate screens.
bash ./scripts/mavros.sh
-
Now run the cams scripts:
screen -S cams python3 -m auv.device.camsVersatile.py
Follow the prompts, enter which camera id you are using and then enter "start" or "stop"
-
If you would like camera feed
screen -S bash cd ~/rtsp ./start_video.sh <camera_id>
-
Run pix_standalone for sensor and thruster capabilities.
screen -S pix python3 auv.device.pix_standalone.py
This will arm the sub by default. To disarm, run
python3 -m auv.utils.disarm, to arm it back again use:python3 -m auv.utils.arm -
Run your mission program
cd auv/missions/ python3 yourMission.py