UBC Snowbots Repository for competitions.
- Installation and Setup
- New Members
- Conventions
- Creating a New Node
- Creating a New Simple Node
- Launch Files
- Testing
- Simulating with Gazebo
- Arduino Development
- Intel Realsense
- Debugging Tips
You will be downloading an Ubuntu ISO and multiple ROS packages with their respective dependencies.
It is highly recommended that you have access to high speed internet while doing this entire setup;
if you're on campus use the ubcsecure or resnet networks for best results.
- Install Ubuntu 18.04 (Note: Download 18.04, not another version, as other Ubuntu versions will make setting up this repo much more painful) (Backup your important data first) (We recommend you use a minimum of 30GB of space)
- For dual-booting: Windows Instructions, Mac Instructions
- If you haven't done so already, setup your UBC alumni email account here
- Using your UBC email account, get a JetBrains education account here
- JetBrains will send an initial email to confirm the UBC email you inputted, once you've confirmed another email will be sent to activate your new education account; you will use this account to set up CLion later on
- Boot into Ubuntu for the remaining steps
- Install git by running
sudo apt-get install git - Fork this repository by heading over to the GitHub page and click the fork button on the top right
- Clone your server-side repository from the terminal by running
git clone --recursive https://github.com/YOUR_USERNAME/Snowflake.git(YOUR_USERNAMEis your github username) - To start the setup run
cd ~/Snowflake && ./setup_scripts/install_tools.sh(Do not run this script as root).- Just choose yes and enter your password when the terminal prompts you
- To build the workspace run
catkin_make. If everything compiles correctly and you don't get any errors, then you're good to go! - If catkin_make fails, and mentions missing .cmake files, run
./setup_scripts/install_dependencies.sh.
- To run CLion with ROS, you must first go in to terminal, navigate to your project (
cd ~/Snowflake), runsource devel/setup.shand then from the same terminal runclion - CLion will not support auto-completion in your .cpp and .h files until you've added them to the CMake file
- Follow the instructions on this github page This repository already contains
zed_ros_wrapperas asubmodulestandalone package insrc/external_pkgs/zed_ros_wrapper(As of 4/27/2018 submodules have been migrated to wstool) - Download ZED calibration file from the link indicated when you run
zed.launchand place it in the folder/usr/local/zed/settings/
Note: External packages are defined as packages which does not have an installable build from apt.
To manually install external packages using wstool, run wstool set src/external_pkgs/<pkg_name> https://github.com/some_org/<pkg_name>.git --git in the workspace root.
The full syntax and options can be found with wstool set -h
To pull in the external package we call wstool update
Note: wstool is addressed and run in install_dependencies.sh, should there be any need to run manual, wstool should be run after all steps from installation and setup has been completed.
Current packages included under wstool:
- Sicktoolbox: Drivers for interacting with Sick Lidars
- Sicktoolbox_wrapper: Interface for Sick Lidar drivers integration into ROS
- ROS_Arduino_bridge: Arduino driver for ROS
- NMEA_Navsat_Driver: NMEA msg driver
- Zed_ROS_Wrapper: Wrapper for interfacing Zed Stereo Cam with ROS
- realsense: Wrapper for interfacing the Realsense D415 with ROS
See here
Tutorials and advice for new members.
First of all, if you're new to the Ubuntu command line: it is primarily used over the graphical user interface as it allows you to do more.
In case you're new to GitHub, get started with this tutorial
Pretty cool right? Now you're wondering how you can use GitHub from the terminal/command line, it's called Git
Now you're wondering how we develop our software, the first place to start is understanding how our workflow works. We use the Forking Workflow, which you can learn more about under Github Conventions in README.md (located in the same directory you opened this readme from) and other important information you should know. If you have questions, you may find your answers over there.
This is the language we use to develop most of our software, so if you are new with C++, understand the fundamentals from this tutorial, but there are much more resources online as well.
By now you understand the basics of ROS, but to learn more about it, you can complete the intermediate tutorials or brush up on the basics if you need to here
Rarely will your Github issues contain all the information that you need to resolve them. Using google to search for that missing piece you need is sometimes useful (think like an optimist when you're understanding/interpreting your project/issue: "If only I knew how to do...", then search "how to do..."), or of course don't hesitate to ask for help, clarification, or just more information from any of the software leads.
Remember it is much better to get help when you're stuck than to waste time trying to figure out what's wrong. This way, you're more or less continuously progressing, and as a whole, we develop efficiently and our software stack progresses.
- We follow the Forking Workflow:
- Only commit files that are essential for the system to run; do not put any photos or videos in here
- All files must be formatted properly. Formatting will be enforced with the
clang-formattool.- To check and fix formatting, from the
Snowflakefolder run./clang_format/fix_formatting.sh BRANCH_NAME, whereBRANCH_NAMEis the name of the branch you intend to merge your code into (ex.iarrcorcore). This script will fix any improperly formatted code, but will refuse to change any files with uncommited changes (to prevent you losing work)
- To check and fix formatting, from the
- Once your pull request has been reviewed and revised until it looks good from both your and the reviewers' sides, go ahead and Squash and Merge it, which will squash all the commits on your pull request into one and merge it to the target branch.
Below are some notes we would like to highlight about pull requests:
- before opening a pull request, quickly run through the pull request checklist and make sure you've satisfied everything
- Once your pull request has been approved, please proceed to merge and close the pull request yourself
- When your pull requests receive comments, please reply to each comment individually.
- If there were any fixes that were specific to resources you found (eg. stackoverflow thread), please comment them into the PR for future reference.
- On a similar note, if you made design decisions, please document them in the comments of the PR. We often go back to close PRs to look at why things were done a certain way. It is very helpful for us to know how you came up with your brilliant solution.
- Every .cpp and .h file should start with
/*
* Created By: Someone
* Created On: December 1st, 2000
* Description: A quick description of what this file does/is for
*/
- Functions should be commented a la JavaDoc
- The Javadoc comment (below) should be directly above every function in the header file
/**
* One line description of the function
*
* A longer and more in depth description of the function
* if it is needed.
*
* @param param_one the first parameter of the function
* @param param_two the second parameter of the function whose
* description goes longer than one line
* @return what the function returns if it returns anything
*
*/
-
Classes are CamelCase
-
Variables are non_camel_case
-
Constants are ALL_CAPS_WITH_UNDERSCORES
-
Functions are camelCase
-
Indentations are 4 spaces
-
CMakeLists.txtfiles should be reduced to only contain the minimum amount of comments. The version insample_packagehas all the comments left in (for the sake of verbosity), so for a more representative example of what yours should look like, seesb_vision/CMakeLists.txt(or really any package aside fromsample_package)
- We try to follow ROS standards, which can be found here
- x : forward
- y : left
- z : up
+X
^
|
+θ +<----->+ -θ
| | |
V | V
+Y <---------------------> -Y
|
|
|
V
-X
- If your node is at all complicated, then this format should be followed. For simple nodes, please see below
- Each node should be class based
- MyNode.h should contain your class declaration
- MyNode.cpp should contain your class definition
- my_node.cpp should be relatively small, and should just contain a main function to run your node
- my-node-test.cpp should contain all your gtest (unit test)
- my_node_rostest.cpp should contain your rostest (integrated test)
- For an example of this, please see
src/sample_package
some_ros_package | CMakeLists.txt | package.xml └───src | | MyNode.cpp | | my_node.cpp | └───launch | | my_node.launch | └───include | | MyNode.h | └───test | | my-node-test.cpp | | my_node_rostest.cpp | | sample_package_test.test
- You will not be able to write unit tests for this type of node, so it must be extremely simple
- my_node.cpp will contain the entire node, and will probably contain a
while(ros::ok()){}loop
some_ros_package | CMakeLists.txt | package.xml └───src | | my_node.cpp
- A launch file (ends in .launch) is an easy way to run multiple nodes with specified parameters with one single command. This is useful if a package has multiple nodes that needs to be run. A launch file can also launch other smaller launch files so it can be used to start every node needed for the robot to run properly. Reference here.
- To use the launch file, the command is:
roslaunch package_name package_launch_file.launch
- GTest is our primary testing tool at the moment. The ROS wiki has a quick intro to it here, and we also strongly recommend you read Google's introduction to it [here] (https://github.com/google/googletest/blob/master/googletest/docs/Primer.md), then setup and write a few example tests before you start using it with ROS.
- Once you've setup your tests in ROS, run
catkin_make run_teststo run them - To test a specific package, run
catkin_make run_tests_MY_PACKAGE_NAME
- For tests which require more than one active node, i.e. integrated testing, the rostest framework provides a way to launch your test alongside all the nodes it requires. This is an extension on roslaunch enabling it to run test nodes. Special test nodes are nested within a
<test></test>tag. This also needs a special entry under CMakelists as shown in the sample package. See more details here - Similar to launch files, the command is:
rostest package_name package_test_file.test.
- You will always need to source the project before running gazebo, by moving to the project directory with
cd ~/Snowflakeand thensource devel/setup.sh - You will probably need a computer with an dedicated gpu, as gazebo sometimes works with intel integrated graphics, but generally not. If you do end up using a computer without a dedicated gpu, make sure to go in to
sb_gazebo/urdf/**ROBOT_NAME**.gazeboand switch around the lidar settings (see comments in said file) - All worlds should go in the
sb_gazebo/worldsfolder - To launch a world, simply run the appropriate launch file in
sb_gazebo/launch - Once the world has launched, it is common for the robot to be initially unable to move. Just lift it up a bit in gazebo and drop it for this to be fixed
- You can manually control the robot with your keyboard, logitech, ps3, or xbox controller. To do so, simply
cd ~/Snowflakeand thensource devel/setup.sh, then run the appriate launchfile fromsrc/sb_gazebo/launchby runningroslaunch LAUNCH_FILE.launch(from within the launch folder). Note: At least with PS3 controllers, you have to hold downL1while using the joystick for the controller to work
- When developing the firmware/Arduino parts of the software, we've made a complete arduino workspace in
src/firmware. This way you don't need to worry about downloading the libraries yourself! - In order to use this, go to your Arduino IDE's Preferences dialog box and use
/your/path/to/Snowflake/src/firmwareas your sketchbook directory. Open arduino sketches in the workspace and they will work!
Intel Realsense is a sensor that captures a 3D depth map of the environment in front of the sensor. This 3D depth map is called a point cloud, which is essentially just a list (vector in c++) that contains a bunch of points, where each point contains the x, y, and z coordinates, as well as a color value associated to it.
- Visualizing the input from the realsense: Run
roslaunch realsense2_camera rs_rgbd.launchto launch realsense. You should be able to see the point cloud at the topic named/camera/depth_registered/pointsinrviz. - Recording a rosbag from the Realsense: Recording a rosbag from the Realsense is useful, because you can record the input from the sensor and play it as many time as you want after recording it. That way, you don't need the physical sensor and you don't to "capture" the same thing every time you run it.
- Launch realsense using
roslaunch realsense2_camera rs_rgbd.launch. - Record a rosbag by running
rosbag record /camera/depth_registered/points <options>. Seerosbag record -hfor the list of options. To stop the recording, just press ctrl+c. E.g.rosbag record /camera/depth_registered/points --split --duration 1m --max-splits 3 -o ../rosbags/tennis_ball/- split files every 1 minute, split for maximum of 3 times, and store the rosbag in../rosbags/tennis_ball/folder. - Stop the realsense node (ctrl+c the process launched in step 1)
- Run
rosbag play path/to/rosbag/file -l.path/to/rosbag/fileis the path to the rosbag generated in step 2. Note that the-lflag loops the recording so it will play forever until you terminate it manually (using ctrl+c). - Open
rviz. You should be able to see the recording of the point cloud at/camera/depth_registered/points.
- Launch realsense using
- If something is happening that does not seem to correspond to your code, (ex. if your node is subscribing to the totally wrong topic) try deleting the
buildanddevelfolders (if present) to remove any "cached" information