ROS$I&interface&for& COMAU&robots&
Transcript of ROS$I&interface&for& COMAU&robots&
INTELLIGENT AUTONOMOUS SYSTEMS LAB
Stefano Michieletto – University of Padua Elisa Tosello – University of Padua Fabrizio Romanelli – Comau SpA Valentina Ferrara – Comau SpA Emanuele Menegatti – University of Padua
ROS-‐I interface for COMAU robots
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
2
About COMAU
CROSS FUNCTIONS
BUSINESS UNITS
INDUSTRIAL OPERATIONS
PROJECT MANAGEMENT OFFICE
ENERGY MANAGEMENT TEAM
BODY WELDING
POWERTRAIN SYSTEMS
ROBOTICS
AEROSPACE
SERVICE
ADAPTIVE SOLUTIONS
QUALITY
• Comau (COnsorzio MAcchine Utensili) is an Italian mulBnaBonal company
• Comau develops and produces process automaBon, manufacturing and service soluBons and specializes in welding robots
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
3
Ini:al COMAU control system
RT UDP/IP
PDL2 Program (Moves, etc.)
C4G Open
C4G Open realistic Robot Library
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
4
RT UDP/IP
PDL2
C4G Open
C4G Open realistic Robot Library
• C4G: Robot Unit Control
• Open: it allows and easy and safe integraBon of the Robot Control Unit with an External
Personal Computer. This helps
• Programming automated roboBc cells
• IntegraBng external sensors to
• ImplemenBng complex manufacturing
applicaBons
Ini:al COMAU control system C4G Open
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
5
Ini:al COMAU control system C4G Open realis:c Robot Library
RT UDP/IP
PDL2
C4G Open
C4G Open realistic Robot Library
ORL allows to perform the following operaBons:
• IniBalizaBon of a virtual robot on Linux (complete Comau robot family) starBng from a real configuraBon file
• ComputaBon of Direct and Inverse KinemaBcs, for each Comau robot
• Error management of posiBon and joint ranges
• IntegraBon of Comau Trajectory Generator and Interpolator
• ComputaBon of the Dynamic Model and Jacobian
The operaBons are realis:c: they describe what the robot would do if they were performed via PDL2 on the
real robot
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
Challenge: • Increase ease of use • Flexibility • Integrability
of the open controller
Solu:on: Development of a ROS-‐I interface for Comau robots
6
Why ROS?
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
7
RT UDP/IP
PDL2 Program (Moves, etc.)
C4G Open
C4G Open realistic Robot Library
ROS
ROS
The final system
ROS
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
8
RT UDP/IP
PDL2 Program (Moves, etc.)
C4G Open
C4G Open realistic Robot Library
ROS
ROS
ROS ROS
The final system
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
9
RT UDP/IP
PDL2 Program (Moves, etc.)
C4G Open
C4G Open realistic Robot Library
ROS
ROS
ROS ROS
TCP/IP
The final system
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
10
RT UDP/IP
PDL2 Program (Moves, etc.)
C4G Open
C4G Open realistic Robot Library
ROS
ROS
ROS ROS
TCP/IP
ROS
ROS
The final system
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
2 ‘hidden’ ROS packages have been developed: -‐ A virtual model of the robot -‐ A plugin integraBng the controller in Gazebo
11
The final system Simula:on
C4G Open realistic Robot Library
ROS
ROS ROS
TCP/IP
ROS
ROS
• Contains the robot meshes • Containts the URDF file
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
2 ‘hidden’ ROS packages have been developed: -‐ A virtual model of the robot
-‐ A plugin integraBng the controller in Gazebo
• Connects the robot controller to the simulator • A wrapper for ORL has been developed • The simulated robot performs the same movementents as the real one
12
The final system Simula:on
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
13
The final system Real robot
RT UDP/IP
C4G Open
ROS ROS
TCP/IP
Problems: -‐ Robot communicaBon → hard real Bme (2 ms) -‐ ROS → not real Bme at all
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
14
The final system Real robot
RT UDP/IP
C4G Open
ROS ROS
TCP/IP
Problems: -‐ Robot communicaBon → hard real Bme (2 ms) -‐ ROS → not real Bme at all
Solu:on: a communicaBon node including
-‐ REAL TIME THREAD → real robot connecBon
-‐ NORMAL PRIORITY THREAD → moBon data
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
The final system is independent from:
-‐ PlaMorm -‐ Both ROS and not ROS –compliant pla^orms can use the so_ware
-‐ Environment -‐ The user can control both a real and a simulated robot -‐ The controller used is the same
-‐ Mo:on planning library -‐ Both ORL and external libraries can provide robot movements
15
The final system Features
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
16
Experiments
-‐ Developed using a Learning From DemonstraBon Framework
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
17
Experiments
-‐ Developed using a Learning From DemonstraBon Framework -‐ MoBon commands sent using ROS -‐ Performed both in the real and in the simulated world -‐ Two scenarios
1. PosiBon controller → ORL 2. Velocity controller → customized MoBon Planning library
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
18
First scenario
Move a box along a linear path from the beginning to the end of a 45 cm long table.
Posi:on controller à ORL
Results of 25 acempts:
45.011 ± 0.402 cm
Simulation Real world
54.332 ± 0.755 cm
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
19
Second scenario
Stop an object in moBon in front of the robot.
Velocity controller à customized MP library
Results of 20 acempts:
Real world
Every Bme the robot stops the object in moBon
INTELLIGENT AUTONOMOUS SYSTEMS LAB
INTELLIGENT AUTONOMOUS SYSTEMS LAB
Now, the COMAU system -‐ Is ROS-‐compliant -‐ Can be used by non ROS-‐compliant pla^orms without integraBons -‐ Lets the user to control both a real and a virtual robot using the same moBon
controller -‐ Allows the replacement of the COMAU MP library with any library of the same
type
20
Conclusions
INTELLIGENT AUTONOMOUS SYSTEMS LAB
21
Thank you for your acenBon