1 of 27

DIGITAL TWINS FOR INDUSTRY 4.0

Presented by Talos Robotics

Hongjian Huo

Ali Entezari

Dmytro Humeniuk

2 of 27

PRESENTATION PLAN

  • Introducing the problem
  • Description of the physical twin
  • Description of the digital twin

3 of 27

4M ROBOTS ARE WORKING AT FACTORIES EVERY DAY

4 of 27

INDUSTRY 4.0 IS THE FUTURE OF MANUFACTURING

5 of 27

SAFETY IS CRITICAL

6 of 27

DIGITAL TWIN: ENABLING SAFETY AND ADVANCED DATA ANALYTICS

7 of 27

DT FOR A PICK AND PLACE ROBOT

8 of 27

PROPOSED DIGITAL TWIN SERVICES AND ARCHITECTURE

9 of 27

��MEET OUR PHYSICAL TWIN:

ROSIE

Kinova Gen3

10 of 27

WE USE A KINOVA GEN3 7-DOF ROBOTIC ARM

11 of 27

PHYSICAL TWIN ARCHITECTURE

Intel NUC

Ubuntu 22.04

Docker container

Workstation

TCP/IP

ROS2 control code

Data out

Data in

DELL ALIENWARE

Ubuntu 22.04

12 of 27

THE ARM IS CONTROLLED VIA ROS2 AND MOVEIT2

Open Motion Planning Library

13 of 27

ROS2 TOPIC DIAGRAM

14 of 27

DT SERVICE 1: SIMULATION

15 of 27

��THE DIGITAL TWIN:

16 of 27

DT ARCHITECTURE

GUI

Communicator

Simulator

Handler

Shared Queue

Shared Queue

PT data

PT data

Stop event

17 of 27

DT SERVICE 2: �VISUALIZATION

  • PyQt5 + Qt designer
    • PyQt Web widget
      • three.js
    • Matplotlib

18 of 27

WE USE LIGHTWEIGHT ROBOTICS SIMULATORS FOR VISUALIZATION AND OFFLINE SIMULATION

19 of 27

DT SERVICE 3: LIGHTWEIGHT SIMULATION

  • Robotics Toolbox Features

- Trajectory planner

    • Controller
    • Swift
    • Rapid prototyping
  • New Added Features
    • online / offline mode
    • Singularity detection
    • Snapshots
  • Easy to Install compared to ROS

Singularity snapshot

20 of 27

DT SERVICE 4:�MONITORING AND EMERGENCY STOP

21 of 27

DT TO PT COMMUNICATION

joints pos:

end-effector pos:

22 of 27

DT CHALLENGES

  • Buggy Swift
    • Bug fix: crashing
    • New feature: Compatibility with PyQt5
    • New feature: Change robot color
    • New feature: Take robot snapshot
    • New feature: having PolyMTL logo :)
  • Connection
    • Delay
  • GUI
    • Synchronization of simulator and plots

23 of 27

PT CHALLENGES

  • Obtaining the correct urdf robot model
  • Planning in Cartesian (3D) space is more challenging than in the joint space
  • Numerous set up issues
  • 3 tickets raised to kinova support, one issue created on kinova github repo

24 of 27

DEMO

Because of upload size limit, the video is removed from the slides.

Video link: DT_DEMO.MOV - Google Drive

25 of 27

CONCLUSIONS

  • We built a DT with 5 services for a Kinova Gen 3 robot
  • Closed loop DT: PT sends data and DT has a control over PT
  • As future work,

26 of 27

ACKNOWLEDGEMNTS

  • We thank Intelligent Cyber-Physical lab for providing access to the robot
  • Big thanks to Ludovic Montagnon, a PhD student in mechanical engineering for his help in the real robot set-up!
  • We thank Prof. Oakes and all course participants for the feedback!

27 of 27

Q & A

Talos Robotics is here to answer questions