1 Ai Kit UI Instructions

AiKit UI Download

  1. Enter URL: https://github.com/elephantrobotics/AiKit_UI/tree/320

  2. download item

    download using git

​ Use git clone, under the 'Code' button, take the first address as an example
28

​ Right click in the folder to open the git window
image-20230109101120225

​ Enter the command:

git clone -b 320 https://github.com/elephantrobotics/AiKit_UI.git

​ Download the ZIP file directly, under the 'Code' button.
image-20230109103257726

start method

path

python3 <path>/main.py

After the startup is successful, as shown in the figure below:

img

Features

language switch

Click the button in the upper right corner of the window to switch between languages (Chinese, English).
img

device connection

  1. Select serial port, device, baud rate
    img
  2. Click the 'CONNECT' button to connect, after the connection is successful, the 'CONNECT' button will change to 'DISCONNECT'
    img

  3. Clicking the 'DISCONNECT' button will disconnect the robot arm
    img

  4. After the robotic arm is successfully connected, the gray button will be lit and become clickable.
    img

Turn on the camera

  1. Set the camera serial number, the default serial number is 0, when Windows is used, the serial number is usually 1, and when Linux is used, the serial number is usually 0.
    img

  2. Click the 'Open' button to try to open the camera. If the opening fails, you should try to change the camera serial number; the camera is successfully opened as shown in the figure below: Note: Before use, the camera should be adjusted to be just above the QR code whiteboard, and there is a line The straight line is facing the mechanical arm.
    img

  3. After successfully opening the camera, click the 'Close' button to close the camera
    img

algorithm control

  1. Fully automatic mode, after clicking the 'Auto Mode' button, the recognition, grabbing, and placing will always be on; click the 'Auto Mode' button again to turn off the fully automatic mode.
    img

  2. Go back to the initial point of grabbing, click the 'Go' button, it will stop the current operation and return to the initial point.
    img

  3. Step-by-step Recognition recognition: click the 'Run' button to start the recognition, Aigorithm is the current algorithm used.
    img Pick: Click the 'Run' button to start the capture. After the capture is successful, the recognition and capture will be automatically closed, and you need to click it again for the next use.
    img Placement: Click the 'Run' button to start placing. The BinA, BinB, BinC, and BinD selection boxes correspond to BinA, BinB, BinC, and BinD 4 storage boxes, respectively, and will be placed in the designated storage box after selection.
    img

  4. Grabbing point adjustment, X offset and Y offset respectively represent the X-axis and Y-axis positions of the mechanical arm coordinates, which can be modified according to actual needs. Click the 'Save' button to save. After saving successfully, it will follow the latest point position to fetch.
    img
    img

  5. Open the file location, our code is open source, you can modify it according to your needs, click the 'Open File' button to open the file location.
    img Open the 'main.py' file and modify it
    img Note: The 'main.py.bak' file is the backup of the 'main.py' file, delete the 'main.py' file when you need to use it, and re-modify the 'main.py. The suffix of the bak' file is 'main.py'; then re-backup the 'main.py' file and name it 'main.py.bak'; you can also choose to re-download the project.

  6. Algorithm selection includes color recognition, color recognition + gripper, shape recognition, QR code recognition, feature point recognition, object recognition, yolov5, intelligent gripper, color recognition + force-controlled gripper, object recognition + force-controlled gripper, and intelligent gripper + force-controlled gripper. Selecting the corresponding algorithm will perform the corresponding recognition.
    img

  7. Add image for feature point recognition or Object recognition, Object recognition Force-controlled gripper Add object.
    img Click the 'Add' button, the camera will open and a prompt will appear.
    img Click the 'Cut' button, the current camera content will be intercepted, and a prompt will be given to 'press the ENTER key after the content needs to be saved'

    img Frame the content to be saved and press the ENTER key to start selecting the saved area, corresponding to BinA, BinB, BinC, BinD 4 storage boxes.

    img The intercepted content will be displayed here
    img

    You can enter the following path to view the saved pictures
    img

  8. Click the 'Exit' button to exit adding pictures. Note: If you start capturing, please exit after capturing. You can choose not to save the captured pictures.
    img

  9. When the algorithm selects Feature Point Recognition, the location placement mode will default to the automatic sorting mode, and the folders where the feature point images are stored will be automatically sorted to the corresponding area positions.
    image-20230106172814972

coordinate display

  1. Real-time coordinate display of the robotic arm: click the 'current coordinates' button to open
    img

  2. Recognition coordinate display: click the ''image coordinates' button to open
    image-20230106180304086

2 Algorithm material corresponding description

Here is a brief explanation of the material usage corresponding to various algorithms:

  • Color recognition-Gripper Model & Suction Pump Model:

color cube

  • Shape recognition-Suction Pump Model:

shape card

  • Feature point recognition-Suction Pump Model:

feature card

  • QR code recognition-Suction Pump Model:

encode card

  • yolov5 recogniton-Suction Pump Model:

yolov5 card

  • Object recognition-Gripper Model:

fruit model

  • Smart gripper-claw type:

Put the AR code on the center of the recognizable object. The AR code can be printed and used by yourself through the download address.

Download address:Smart_gripping_AR_code_materials.pdf

smart model

3 Case Operation Demonstration

1 Suction Pump Model

1.1 Color Recognition

1.2 Shape Recognition

1.3 AR QR code recognition

1.4 Feature point recognition

1.5 Yolov5 recognition

2-claw model

2.1 Color recognition-adaptive gripper

2.2 Object Recognition - Adaptive Gripper

2.3 Intelligent gripping - adaptive gripper

2.4 myGripper F100 force-controlled gripper

Supports color recognition, object recognition, and intelligent gripping.

results matching ""

    No results matching ""