Formula Student Driverless - developed by University of Gyor

Proposed system architecture

The proposed system architecture is shown below. The following considerations are made for each component:

• state machine: base logic of szenergy bringup is taken over, with slight modifications to FS rules; output topic includes both state and mission ID, and mission ID is set manually (external source or parameter in accordance with FS rules)
• object detection: two parallel solutions are implemented, later either of these or the combination (fusion) is used, to be decided later,
• object fusion: if e.g., only lidar detection is used, fusion is unnecessary and can be deleted,
• map creation includes SLAM
• trajectory planner receives the raw object list and plans path based on it (e.g., first round of trackdriver, acceleration) and also received the map from the slam, where global trajectory is planned; this switching is handled inside the planner,
• control in the first place is a simple take-over of the szenergy solution, later it is planned to be changed to handle high speed handling.

All components are to be developed in ROS 2 Humble.

flowchart TD
    S[State Machine <br>/plan_state_machine] -.->|/plan_state*| LP[LIDAR pre filter<br>/prcp_lidar_filtered]
    LP --> LS[LIDAR segementation<br>/prcp_ground_obstacle_segm_lidar]
    S -.-> CS[Cone detection camera<br> and de-projection]
    S -.-> O[Object fusion]
    CS -->|/prcp_obj_list_camera| O
    LS -->|/prcp_obj_list_lidar| O
    O -->|/prcp_obj_list_fused| T[Trajectory planner<br>/plan_trajectory_planner]
    T --> C[Control<br>/ctrl_vehicle_control]
    S -.-> T
    S -.-> C
    O --> M[Map Creation<br>/prc_slam]
    M -->|/prcp_map| T
    L[Localization<br>/prcp_odometry_kf_prediction] --> T
    C --> CAN[To CAN]
    classDef light fill:#34aec5,stroke:#152742,stroke-width:2px,color:#152742  
    classDef dark fill:#152742,stroke:#34aec5,stroke-width:2px,color:#34aec5
    classDef white fill:#ffffff,stroke:#152742,stroke-width:2px,color:#15274
    classDef dash fill:#ffffff,stroke:#152742,stroke-width:2px,color:#15274, stroke-dasharray: 5 5
    classDef red fill:#ef4638,stroke:#152742,stroke-width:2px,color:#fff
    class CS,LS,LP,L,T,M,C white
    class O light
    class S dash
    class CAN red

Loading

Clone and build

Note

There is a dockerfile included in order to launch the whole system containerized. ( instructions below + more details in corresponding subdirectory: /docker/README.md )

Clone

cd ~/ros2_ws/src

git clone https://github.com/szenergy/formula_student_packages

Build _{^(non-Docker)}

sudo apt install ros-humble-pcl-ros

cd ~/ros2_ws 

colcon build --symlink-install --packages-select formula_student_bringup cone_detection_lidar cone_detection_camera lidar_pre_filter

Docker build

( from the same directory as the dockerfile, assuming Git is installed )

git clone https://github.com/jkk-research/mmdetection3d/

docker build -t formula_student_stack_image .

Run _{^(non-Docker)}

source ~/ros2_ws/install/setup.bash 

E.g:

ros2 launch formula_student_bringup tf_static.launch.py

ros2 launch cone_detection_lidar detection_simple.launch.py

Docker run

Native Linux run command:
To be able to open a window from within the container, run this first: xhost +local:docker (host machine window access)

docker run -it --rm --net=host -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix -v $HOME/.Xauthority:/root/.Xauthority:rw --name formula_student_stack formula_student_stack_image

WSL (Windows) run command:

docker run -it --rm --net=host -e DISPLAY=host.docker.internal:0 -e LIBGL_ALWAYS_INDIRECT=0 -e XDG_RUNTIME_DIR="$XDG_RUNTIME_DIR" -v /tmp/.X11-unix:/tmp/.X11-unix --privileged --gpus all --name formula_student_stack formula_student_stack_image

Directory structure

Caution

If you move the packages to another directory, you need delete the build and install directories and rebuild the packages.

~/ros2_ws/src/
    ├── formula_student_packages # this repo
    │   ├── cone_detection_lidar
    │   │   ├── launch
    │   │   └── src
    │   ├── formula_student_bringup
    │   │   └── launch
    │   ├── other fs package1
    │   ├── other fs package2
    │   └── img
    └── wayp_plan_tools # other non-fs packages
        ├── csv
        ├── launch
        └── src

Related packages

• https://github.com/url-kaist/patchwork-plusplus-ros/tree/ROS2
• https://github.com/jkk-research/lidar_cluster_ros2
• https://github.com/jkk-research/pointcloud_to_grid