Digital twin for adaptation of robots’ behavior · Mobile Robot Platform (MRP) “Enabling...
Transcript of Digital twin for adaptation of robots’ behavior · Mobile Robot Platform (MRP) “Enabling...
CIRP CARV Conference 2018 08-10 October 2018
Nantes, France
Digital twin for adaptation of robots’ behavior in flexible robotic assembly lines
Niki Kousi, Christos Gkournelos, Sotiris Aivaliotis, Christos Giannoulis, George Michalos, Sotiris Makris
Laboratory for Manufacturing Systems & Automation (LMS)Dept. of Mechanical Engineering & AeronauticsUniversity of Patras, Greecewww.lms.mech.upatras.gr
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
2
Contents
▪ Introduction
▪ Hybrid production paradigm
▪ Digital twin infrastructure
• Resource Manager
• Sensor manager
• 3D Environment constructor
▪ Unified data model
▪ Automotive case study
▪ Conclusions
▪ Acknowledgments
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
3
Introduction
Today – Tradition production Tomorrow – Future production
Production system transition to mass customization
Stationary resources – rigid line structure
Fixed control logic of the process
High cost – effort to make changes in the line-process
Market demand for higher product personalization
Need for multi product variants
Need for Flexible and reconfigurable production lines
Lead to
combined with Lead to
Lead to
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
4
Hybrid production paradigm
How to enable the dynamic nature of this production paradigm ?
“Perception & skills to
automatically program and
execute multiple tasks”
“Human-Robot &
Robot-Robot
collaboration in a safe
way”
Mobile Product Platform (MPP)
Mobile Robot Platform (MRP)
“Enabling mobility on
products and resources”
Product & Process Perception
Environment Perception “Dynamic balancing and
redirecting to stations”
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
5
Digital twin infrastructure
Resource Nodes
Resource
Manager
Sensor
Manager
Layout
Manager
3D environment
constructor
Unified
Data
Model
Sensor Nodes
3D Scene reconstructionArm Motion
Planner
Platform
Path Planner
Robot behavior
adaption
Digital Twin
CAD files
Execution Coordinator
CAD
Sensors
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
6
Resource Manager
ROS Driver Driver
: Resource manager
MRP_1 Status topic
MRP_1 Location topic
MPP_1 Status topic
MPP_1 Location Topic
MPP_1MRP_1
: Status Monitoring
: Location Monitoring
: Status Monitoring
: Location Monitoring
ROS Topic Publisher
ROS Topic Publisher
ROS Topic Subscriber
ROS Topic Subscriber
ROS Topic Subscriber
ROS Topic Subscriber
• The registration of a new
resource into the production
system
• The storage of the appropriate
attributes based on the unified
resource model
• Realtime monitoring
❖ Resource location
❖ Resource execution status
The responsibilities of resource
manager are:
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
7
Sensor Manager
ROS
Driver
rc_visard 65KinectMicroScan3rc_visard 160
ROS
Driver
ROS
Driver
ROS
Driver
ROS Topic Publisher
ROS Topic Publisher
ROS Topic Publisher
ROS Topic Publisher
: Sensor manager
rc_visard 160 topic
MicroScan3 topic
Kinect topicrc_visard 65 topic
: 2D-3D Sensor Data combination
Fused data topic
ROS Topic Subscriber
ROS Topic Subscriber
ROS Topic Subscriber
ROS Topic Subscriber
ROS Topic Subscriber
• The interfacing with the existing’s
sensors’ ROS drivers.
• The storage of the appropriate
configuration files based on the
unified sensor model
• Sensor data combination
for merging multiple, sensor data
into one topic
The responsibilities of sensor
manager are:
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
8
Layout Manager
• The storage of the CADs files of
fixtures, parts and products.
• The creation of the .sdf files
containing the collisions and the
inertia attributes
• The relative positions between
parts and static fixtures
The responsibilities of layout manager are:
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
9
3D Environment Constructor
Generated Tf tree
3D Environment constructor
rc_visard URDF file
MRP URDF file
Product SDF file
TF topic tree
MPP URDF file
Kinect URDF file Laser Scanner URDF file
Fixtures SDF file
Sensors Manager
Layout s Manager
Re
sou
rce
s M
ana
ge
r
Publish tf data
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
10
Unified world data model
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
11
Automotive case study (1/3)
Station Task Resource Duration
S1 Pre-assembly of right damper Human 12 secs
S1 Load damper on comp. machine Human 5 secs
S1
Compression of right damper & Alignment / Nut insertion during the compression
Human –Comp. mach.
20 secs (human -8 secs)
S1 Load comp. damper on AGV Human 5 secs
S2 Pre-assembly of left damper Human 12 secs
S2 Load damper on comp. machine Human 5 secs
S2 Compression of right left & Alignment / Nut insertion during the compression
Human –Comp. mach.
20 secs (human -8 secs)
S2 Load comp. damper on AGV Human 5 secs
S3 Screwing machine connect each damper with one disk
Screwing machine
57 secs
S4 Cables / Screws Insertion Human 50 secs
• Assembly of the front suspension of a
passenger vehicle
• Involves handling, cabling insertion and
screwing operations – 4 workstations
• Manual assembly & 1 automated screwing
machine
• Challenges of the current set up
❖ Ergonomy: up to 480 dampers (6 Kg)
are lifted by human operators per shift
❖ Flexibility: the automated screwing
machine cannot handle all product
variants
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
12
Automotive case study (2/3)
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
13
Automotive case study (3/3)
MRP H/W configuration Online robot arm motion planning
3D Octomap
Online robot base path planning
Moving
obstacle
Integrated software components:
Online robot instruction with gestures
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
14
Conclusions
The benefits of the proposed infrastructure for a production
system
✓ Robustness against the real-world uncertainty
✓ Generic enough to be applicable in various use cases
✓ Fast and easy integration of new resources
Future work
➢ Integration with the physical robotic set up
➢ Performance validation in the proposed infrastructure
➢ Integration with a high level decision mechanism for online
task re-allocation
Laboratory for Manufacturing Systems and Automation (LMS)Department of Mechanical Engineering and Aeronautics, University of Patras, Greece
15
Acknowledgments
THOMAS
Mobile dual arm robotic workers with embedded cognition for
hybrid and dynamically reconfigurable manufacturing systems(Grant Agreement: 723616)
http://www.thomas-project.eu
http://www.thomas-project.eu/facebook
http://www.thomas-project.eu/youtube
http://www.thomas-project.eu/twitter
http://www.thomas-project.eu/googleplus
http://www.thomas-project.eu/linkedin
CIRP CARV Conference 2018 08-10 October 2018
Nantes, France
THANK YOU!
Laboratory for Manufacturing Systems & Automation (LMS)Dept. of Mechanical Engineering & Aeronautics
University of Patras, Greecewww.lms.mech.upatras.gr