Setting Up ROS Environment On The Raspberry PI

ROS stands for Robot Operating System, which modularizes your code and provide a template for communication across different parts of the robot (different parts of your code). More about ROS can be found on their official guide.

It's recommended to setup the ROS environment inside a Docker container. A container is similar to a Virtual Machine(VM) but is more lightweight. More about docker can be found on their website.

Yahboom's Raspbot V2 comes with some premade ROS nodes which act as an interface to code around it. Its usage is documented on their official guide under 9. ROS2-Robot chassis control course and 10.ROS2-OpenCV course.

The easiest way to setup the ROS enviroment in a docker container is to flash the default .img which can be found on their website under System_File which redirects you here.

Otherwise, it's possible to build a custom Docker image from a custom Dockerfile to setup the environment. The source files for their ROS workspace can be found in their website under Code which redirects here.

.
├── instruction_manual
├── project_demo
│   ├── 03.Basic_car_course
│   │   └── RGB_Light_bar_test
│   ├── 04.Car_motion_control
│   │   └── __pycache__
│   ├── 05.Advance_course
│   ├── 06.Open_source_cv_course
│   │   ├── A.introduction
│   │   ├── B.Geometric_Transformations
│   │   ├── C.Image_Processing_Text_Drawing
│   │   ├── D.Image_Enhancement
│   │   └── E.Machine_Learning
│   ├── 07.AI_Visual_Recognition
│   │   ├── 01.Camera_Driving
│   │   ├── 02.Color_Recog
│   │   ├── 04.Tensorflow_object_recognition
│   │   ├── 05.QR_code_recognition
│   │   ├── 06.Face_recognition
│   │   ├── 07.License_plate_recognition
│   │   ├── detection
│   │   └── mediapipe
│   ├── 08.AI_Visual_Interaction_Course
│   │   ├── 01.Color_tracking
│   │   ├── 02.Color_follows
│   │   ├── 03.QRcode_recog_campaign
│   │   ├── 04.Face_tracking
│   │   ├── 05.Face_follow
│   │   ├── 06.Vision_Based_Auto_LineFollowing
│   │   ├── 07.Garbage_identification_yolov4tiny
│   │   ├── 07.Garbage_identification_yolov5lite
│   │   ├── 08.Autopilot_map_sandbox
│   │   └── 09.Gesture_follows
│   ├── lib
│   │   └── __pycache__
│   └── raspbot
│       ├── data
│       ├── object_detection
│       ├── __pycache__
│       ├── ssdlite_mobilenet_v2_coco_2018_05_09
│       └── templates
└── ros2_source_code
    ├── library_ws
    │   ├── bin
    │   ├── include
    │   ├── lib
    │   ├── lib64 -> lib
    │   ├── py_install
    │   ├── software
    │   └── test
    ├── yahboomcar_ros2_ws
    │   └── yahboomcar_ws
    └── yahboomcar_ws
        └── src

54 directories

The src directory for the workspace is stored at ./ros2_source_code/yahboomcar_ws/src.

An example setup may look something like this:

.
├── Dockerfile
├── docker_ros2.sh
└── src
    ├── yahboomcar_apriltag
    ├── yahboomcar_astra
    ├── yahboomcar_bringup
    ├── yahboomcar_ctrl
    ├── yahboomcar_description
    ├── yahboomcar_mediapipe
    ├── yahboomcar_msgs
    ├── yahboomcar_point
    └── yahboomcar_visual

In the Dockerfile:

FROM ros@sha256:5ebfd9b9282d540dadcb3affe02d90496b075a8e81c70b235a1e56e0788d610a

# Install required system packages
RUN apt-get update && apt-get install -y \
    ros-humble-cv-bridge \
    ros-humble-usb-cam\
    python3-smbus \
    python3-pip \
    libgl1-mesa-glx \
    micro 

RUN sudo apt install -y libxcb-xinerama0 libxcb-xinerama0-dev libxcb1 libx11-xcb1 libxcb-render0 libxcb-shm0

RUN apt install -y libpcl-dev

RUN rm -rf /var/lib/apt/lists/*

# Install Python packages
RUN pip3 install mediapipe opencv-python

RUN echo "source /ros_entrypoint.sh" >> .bashrc

WORKDIR /ros2_ws

COPY . .

CMD ["/bin/bash"]

Inside docker_ros2.sh:

#!/bin/bash
xhost +
docker run -it \
--privileged=true \
--env="DISPLAY" \
--env="QT_X11_NO_MITSHM=1" \
-v /tmp/.X11-unix:/tmp/.X11-unix \
--security-opt apparmor:unconfined \
-v /home/pi/temp:/root/temp \
-v /dev/i2c-1:/dev/i2c-1 \
-v /dev/i2c-0:/dev/i2c-0 \
--device=/dev/video0 \
--device=/dev/video1 \
--device=/dev/gpiomem \
ros_for_raspbot_custom:latest /bin/bash

Build the environment using Docker

docker build -t ros_for_raspbot_custom:latest

Run the docker environment by running the docker_ros2.sh script

bash ./docker_ros2.sh

Once inside the Docker environment, build the ROS workspace using colcon

colcon build

Source the setup.bash.

source ./install/setup.bash

It's then possible to follow the tutorials in the official website.

Common Problems and its fix

• ros2 command not found can be fixed by sourcing /ros_entrypoint.sh
• not being able to access Yahboom's premade modules can be fixed by sourcing ./install/setup.bash
• GUI not being able to launch can be fixed by first running bash ./docker_ros2sh while using remote desktop, such that echo $DISPLAY gives :0.
• the apriltag modules generally don't work due to np.float being deprecated in numpy version 1.20, can be manually fixed by changing np.float to just float