The motion estimation and mapping team seeks to localize the position of the car by fusing data from sensors such as the GPS, IMU, wheel speed sensor (and others) with the mapping from LIDAR sensors. In this new member project, you will learn the basics of ROS2 and using the robot_localization package for sensor fusion.
Start by installing ROS2 Galactic and read through the docs to learn the basics of nodes, topics, services, rosbags etc. (Don't need to cover all of it, just could learn as you go along).
ROS2: https://docs.ros.org/en/galactic/Installation.html#installationguide
Note: Ros2 Galactic is the latest version of Ros2. It is backwards compatible, so it is completely fine if you have earlier version of Ros2 e.g. Ros2-Foxy
The task is to use the robot_localization package (which implements an EKF for you already) and to set it up with a provided rosbag2 (a collection of data) and to output the filtered data. You can install robot_localization package after installing ROS2 by running the command:
sudo apt install ros-[ros-version]-robot-localization replace [ros-version] with galactic or foxy or whatever version of ROS2 you're using.
ROSBAG Download Link (download the entire folder): https://drive.google.com/drive/folders/1pEx0RrZlvdcZNCdEVQhpvi4UzvTHB7Y5?usp=sharing
The message types that are provided in the rosbag is
- sensor_msgs/Imu (IMU data)
- geometry_msgs/TwistWithCovarianceStamped (Optical Speed Sensor)
- sensor_msgs/NavSatFix (GPS)
A step by step tutorial including documentation (which fuses different sensors) is provided in the useful references at the bottom. Reading through these should help you setup the package properly with configuration files.
Its important to note you're not implementing EKF yourself (there is a ROS2 node already that does it for you http://docs.ros.org/en/noetic/api/robot_localization/html/state_estimation_nodes.html). You're just configuring this ROS2 node such that it works with the sensor data thats provided to you in a rosbag. A sample configuration file is provided here https://github.com/cra-ros-pkg/robot_localization/blob/foxy-devel/params/ekf.yaml which sets up this ros2 ekf node (which is documented in the link above).
All your project files including configuration files, launch files, any code you've written, and steps to run them and view results if you've added anything different.
robot_localization docs: http://docs.ros.org/en/melodic/api/robot_localization/html/index.html#
Useful tutorial covering similar concepts: https://automaticaddison.com/sensor-fusion-using-the-robot-localization-package-ros-2/
ROS2 Docs
- Configuring ROS environment: https://docs.ros.org/en/galactic/Tutorials/Configuring-ROS2-Environment.html
Just understanding the below concepts and the commands to interact with them should be enough (but feel free to execute them on the turtlesim if you want).
- ROS2 Nodes: https://docs.ros.org/en/galactic/Tutorials/Understanding-ROS2-Nodes.html
- ROS2 Topics: https://docs.ros.org/en/galactic/Tutorials/Topics/Understanding-ROS2-Topics.html
- ROS2 Param: https://docs.ros.org/en/galactic/Tutorials/Parameters/Understanding-ROS2-Parameters.html
- ROS2 Launch Files (not necessary but could make life easier so you don't have to enter a million commands): https://docs.ros.org/en/galactic/Tutorials/Launch-Files/Creating-Launch-Files.html
Param files are how you're going to be setting up this ROS2 node. These are .yaml files that allow you to customise how the node functions. This is a pretty important bit for this project so here's an another tutorial explaining them in a bit more depth. Its also explained in the sensor fusion tutorial above
https://roboticsbackend.com/ros2-yaml-params/
The parameters that can be modified on ekf_localization_node can be found here http://docs.ros.org/en/melodic/api/robot_localization/html/state_estimation_nodes.html
A sample configuration file for this node is here: https://github.com/cra-ros-pkg/robot_localization/blob/foxy-devel/params/ekf.yaml