qwenvl2_car_recorder.yml

nodes:
  - id: mickrobot_chassis # Run on car
    # build: wget -nc https://github.com/dora-rs/gosim-2024/releases/download/v0.0.1/robot # cargo install --git https://github.com/lyon-dora/chongyou-car
    path: robot
    _unstable_deploy:
      machine: ec7
    inputs:
      text: dora-qwenvl-recorder/text

  - id: keyboard-listener # Run on car
    # build: pip install dora-keyboard
    path: dora-keyboard
    inputs:
      tick: dora/timer/millis/10
    outputs:
      - char
    env:
      DISPLAY: :1

  - id: key-interpolation
    build: wget -nc https://raw.githubusercontent.com/lyon-dora/dora_mickrobot/refs/heads/main/key_interpolation.py
    path: key_interpolation.py
    inputs:
      keyboard: keyboard-listener/char
    outputs:
      - text

  - id: camera1 # Run on car
    # build: pip install opencv-video-capture
    path: opencv-video-capture
    _unstable_deploy:
      machine: ec7
    inputs:
      tick: dora/timer/millis/200
    outputs:
      - image
    env:
      CAPTURE_PATH: 0
      IMAGE_WIDTH: 640
      IMAGE_HEIGHT: 480
      ENCODING: jpeg
      FLIP: VERTICAL

  - id: camera2 # Run on car
    # build: pip install opencv-video-capture
    path: opencv-video-capture
    _unstable_deploy:
      machine: ec7
    inputs:
      tick: dora/timer/millis/200
    outputs:
      - image
    env:
      CAPTURE_PATH: 2
      IMAGE_WIDTH: 640
      IMAGE_HEIGHT: 480
      ENCODING: jpeg
      FLIP: VERTICAL

  - id: plot
    # build: |
    # wget -nc https://github.com/dora-rs/gosim-2024/releases/download/v0.0.1/rerun
    # wget -nc https://github.com/dora-rs/gosim-2024/releases/download/v0.0.1/dora-rerun
    path: dora-rerun
    inputs:
      image_1:
        source: camera1/image
        queue_size: 1
      image_2:
        source: camera2/image
        queue_size: 1
      textlog_vlm: dora-qwenvl-recorder/text
    env:
      IMAGE_WIDTH: 640
      IMAGE_HEIGHT: 480
      RERUN_MEMORY_LIMIT: 25%
      README: |
        To control the robot, use the following keys:
        - w: move forward
        - s: move backward
        - a: move left
        - d: move right
        - q: stop
        - r: open gripper
        - t: close gripper
        - f: move arm left
        - g: move arm right
        - h: move arm down
        - y: move arm up
        - v: go home
        - b: save home

  - id: dora-qwenvl-recorder
    build: |
      git clone https://github.com/hiyouga/LLaMA-Factory LLaMA-Factory
      pip install llama-factory-recorder
    path: llama-factory-recorder
    inputs:
      image_1:
        source: camera1/image
        queue_size: 1
      image_2:
        source: camera2/image
        queue_size: 1
      ground_truth: key-interpolation/text
    outputs:
      - text
    env:
      DEFAULT_QUESTION: Respond with right, left, forward, backward,stop to move to the red line
      LLAMA_FACTORY_ROOT_PATH: LLaMA-Factory