08-19-2026, 12:14 PM
(This post was last modified: 08-19-2026, 12:21 PM by Muhammad Faizan Ahmed.)
Hello everyone,
I am currently working on a project in RoboDK involving two JAKA collaborative robots (Zu7 and Zu12). I am using a Python script in RoboDK to control the robots and integrate a real-time vision system using a USB webcam and a YOLO11n model.
My robot movement program works correctly and the physical robots move as expected when I run the script without the YOLO/Ultralytics import.
However, when I add:
the RoboDK simulation continues to work, but the physical robots do not move. Interestingly, if I comment out the Ultralytics import, the physical robot movement works again.
I am using:
My intended workflow is:
Any guidance on how to diagnose this issue would be greatly appreciated. See the attached code file for reference.
Thank you.
I am currently working on a project in RoboDK involving two JAKA collaborative robots (Zu7 and Zu12). I am using a Python script in RoboDK to control the robots and integrate a real-time vision system using a USB webcam and a YOLO11n model.
My robot movement program works correctly and the physical robots move as expected when I run the script without the YOLO/Ultralytics import.
However, when I add:
Code:
from ultralytics import YOLOthe RoboDK simulation continues to work, but the physical robots do not move. Interestingly, if I comment out the Ultralytics import, the physical robot movement works again.
I am using:
- RoboDK
- Python 3.10.11
- Ultralytics
- OpenCV
- YOLO11n
- JAKA Zu7 and Zu12
My intended workflow is:
- Move the robots to predefined positions using RoboDK.
- Start the webcam-based YOLO11n vision system.
- Detect the needle and gripper states.
- Based on the detection, continue the robot movement sequence.
- Execute the same sequence on the physical JAKA robots.
- Could the Ultralytics/PyTorch libraries interfere with RoboDK's physical robot driver, Python execution, or communication with the robot?
- Is there a recommended way to run YOLO/OpenCV vision processing alongside RoboDK's physical robot control without preventing the robot driver from executing?
- Would it be better to run the vision processing in a separate Python process/thread and communicate the detection result to the RoboDK script?
Any guidance on how to diagnose this issue would be greatly appreciated. See the attached code file for reference.
Thank you.

