Kalman's software stack updated to ROS 2
Note
It is strongly recommended to use kalman_ws as the workspace for this project. It is a prebuilt ROS 2 workspace that contains all the necessary dependencies for this project. It also includes useful shell macros to facilitate the installation of custom Python and APT packages. This workspace requires Ubuntu Jammy, but it also fully supports execution and development in a Docker container via Distrobox.
- Spatio-Temporal Voxel Layer (Currently not available via rosdep and has to be built manually from source.) Required by
kalman_nav2
. - the latest ROS 2 Rolling version of the realsense-ros package which supports way more features and has a working (!) auto-exposure feature
- the latest commit of rosbridge_suite which has support for ROS 2 actions. It is used in
kalman_gs
package. - our custom version of compressed_depth_image_transport that supports RVL compression and does not clash with
compressed_image_transport
's params - our custom version of rtabmap_odom that supports compressed image transports
- latest version of the ros2 branch of moveit_msgs
- All Python packages listed in the
requirements.txt
files in the directories of some Colcon packages. - All APT packages listed in the
apt_packages.txt
files in the directories of some Colcon packages.
To make development quicker, the launch process of the development setup has been split into two separate launch files. One of them, suffixed _sim_base
, launches the simulation environment and RViz. The other one, suffixed _sim_stack
, launches the rest of the software stack, including localization, mapping, and navigation capabilities:
ros2 launch kalman_bringup urc_autonomy_sim_base.launch.py
# Before running the next command, make sure that the simulation is started and sensor messages are being published.
ros2 launch kalman_bringup urc_autonomy_sim_stack.launch.py
Note
The commands shown above will launch the Unity simulation environment which requires additional installation. Please follow the instructions here in order to set up Unity.
In order to run on a physical rover, just one launch file is needed:
ros2 launch kalman_bringup urc_autonomy_rover.launch.py
To launch the ground station, use the following command on a separate machine:
ros2 launch kalman_bringup urc_autonomy_gs.launch.py
For certain features to work, the ground station should be connected to the same network as the rover and/or have our custom radio communication hardware connected to it.
Multiple other launch files are available in the kalman_bringup
package, including the ones for other competitions and development purposes.
Kalman's software stack is composed of multiple packages that are meant to be built and run together:
kalman_aruco
- ArUco tags detection using aruco_opencvkalman_bringup
- launch files for the rover and the ground stationkalman_clouds
- point cloud generation and filteringkalman_description
- Xacro / URDF descriptions + models for the roverkalman_drivers
- drivers for the physical hardware; only to be run separately from the simulationkalman_gazebo
- Ignition Gazebo simulation configs for Kalmankalman_interfaces
- ROS 2 messages, services and actions used by the otherkalman_
packageskalman_mapviz
- configuration and launch files for MapVizkalman_master
- drivers for our custom Master devicekalman_nav2
- configuration and launch files for Nav2 and related modules; Includes a custom path follower.kalman_robot
- a metapackage that depends on all otherkalman_
packageskalman_slam
- configuration files for robot_localization and RTAB-Mapkalman_supervisor
- Manages autonomous navigation missions.kalman_wheels
- a node that converts Twist messages on/cmd_vel
and similar topics to the actual wheel state; Also includes safeguards that can limit the acceleration and velocity or stop the rover to adjust wheel rotation.kalman_yolo
- PRIVATE models and configs foryolo_ros
point_cloud_utils
- utilities for working with point clouds; Includes ROS wrappers around PCL filters and an obstacle detection node.service_based_nav2_controller
- aFollowPath
controller plugin for Nav2 that uses a service to compute velocity commandsunity_sim
- a Unity-based simulation environment that can seamlessly replace the physical hardware of AGH Space Systems' robotsyolo_ros
- YOLO-based object detector; Supports composition and lifecycle management.
Launch files are organized in a hierarchical manner. The kalman_bringup
package contains main launch files that are meant to be the only ones used via ros2 launch
. kalman_bringup
includes many other launch files from other kalman_
packages, which in turn may include even more launch files from other packages:
All kalman_
packages are designed to work together and exchange data in a complex manner. The following diagram shows a high-level overview of the data flow between top-level modules:
As mentioned in Launch Hierarchy, top-level modules may include other modules that are not shown in the diagram. Each module contains a set of nodes that actually perform the data processing and exchange.
- When committing new code, please follow the Conventional Commits specification.
- The
scripts
directory should only contain scripts that are meant to be installed as ROS executables. - The
tools
directory should only contain developer utilities and not ROS-related code. - Whenever possible, design your C++ nodes as components. See this tutorial for more information.
- If a dependency is not available in rosdep, please add it to the
apt_packages.txt
orrequirements.txt
file created next topackage.xml
. Always prefer rosdep over those files.
- Nav2 (STVL) on random occasions when booting up, starts spamming TF errors and takes up 100% of the CPU. Issue is present both in simulation and on the physical rover.