Prototyping Plant ControlSoftwarewith with Microsoft Robotics Studio Third International Workshop on Software Development and Integration in Robotics Alwin Hoffmann, Florian Nafz, Frank Ortmeier, Andreas Schierl, Wolfgang Reif Department of Software Engineering and Programming Languages University of Augsburg Germany
Agenda A short vision ii of tomorrow s industrial i robotics Case study: an adaptive production cell How can Microsoft Robotics Studio support us? The adaptive production cell implementation in Microsoft Robotics Studio Problems & Open Issues Future Work Conclusion May 2008 Hoffmann et.al., University of Augsburg, Germany 2
Challenges of Industrial Robotics Quantity Tasks Flexibility large-volume usually limited variability production standard tasks May 2008 Hoffmann et.al., University of Augsburg, Germany 3
Challenges of Industrial Robotics Until now: large volume production standard tasks In the future: low volume production easy reconfiguration short installation time special ilapplications i short development times new domainsfor industrial robots flexible use of industrial robots easy adoption to new production processes limited variability May 2008 Hoffmann et.al., University of Augsburg, Germany 4
Requirements Today we have staticcontrol control structures Future production systems should be reconfigurable fault tolerant suitable for different (similar) tasks Organic Computing systems meet these requirements adaptive systems with self X properties (self configuration, self healing, self adapting, ) a.k.a. autonomic/autonomous computing for the design and construction of organic computing systems see [Seebach et.al. 2007] May 2008 Hoffmann et.al., University of Augsburg, Germany 5
Traditional production cell Goal: Process in a given order! unprocessed processed Traditional production systems Adding a new robot can be done only manually The failure of one component causes the failure of the whole system Introducing new requires changing the controller software May 2008 Hoffmann et.al., University of Augsburg, Germany 6
Case Study: Adaptive production cell Whatis different when using an adaptive production cell? Flexible hardware components Flexible industrial robots with multiple tools Flexible transport system (autonomous carts) Workpieces with RFID tags Planning component Monitoring of the components and the Allocation of tasks by (re )configuration of the components May 2008 Hoffmann et.al., University of Augsburg, Germany 7
Case Study: Adaptive production cell unprocessed processed May 2008 Hoffmann et.al., University of Augsburg, Germany 8
Case Study: Adaptive production cell Goal: Adding a new robot! unprocessed processed May 2008 Hoffmann et.al., University of Augsburg, Germany 9
Case Study: Adaptive production cell Goal: Adding a new robot! unprocessed processed Self configuration: o : dynamically y integrate new robots May 2008 Hoffmann et.al., University of Augsburg, Germany 10
Case Study: Adaptive production cell Problem: Fil Failure of a single component! unprocessed processed May 2008 Hoffmann et.al., University of Augsburg, Germany 11
Case Study: Adaptive production cell Problem: Fil Failure of a single component! unprocessed My drill is broken so I switch to tighten screw I calculated the new transportation ordering So I can stay in insert screw processed Okay, I switch to drill hole So let s change to the new order Self healing: ea compensate to component failures May 2008 Hoffmann et.al., University of Augsburg, Germany 12
Case Study: Adaptive production cell Goal: Adopt to new work plans! Partly processed and unprocessed (with RFID tags) unprocessed processed May 2008 Hoffmann et.al., University of Augsburg, Germany 13
Case Study: Adaptive production cell Goal: Adopt to new work plans! Partly processed and unprocessed (with RFID tags) unprocessed processed Self adapting: Adopt to new goals May 2008 Hoffmann et.al., University of Augsburg, Germany 14
Case Study: Adaptive production cell Sounds nice But does it work? May 2008 Hoffmann et.al., University of Augsburg, Germany 15
What is Microsoft Robotics Studio? Windows based development environment for robotics applications Asynchronous, service oriented runtime Visual programming environment 3D physics based simulation environment Supports a variety of hardware platforms Concurrency and Coordination Runtime (CCR) Concurrent, message based programming model Asynchronous operations Decentralized Software Services (DSS) Distributed & service oriented application model Hosting environment for services May 2008 Hoffmann et.al., University of Augsburg, Germany 16
How can Microsoft Robotics Studio support us? Microsoft ftrobotics Studio facilitates t the fast prototyping of novel robotics applications by providing an asynchronous, service oriented architecture a realistic physics based simulation engine Benefits: Complex applications are possible Composition of services Reuse of services No need of real hardware Proof of concept Lower development costs May 2008 Hoffmann et.al., University of Augsburg, Germany 17
Implementation The industrial robots Based on the simulated KUKA LWR3 arm from the KUKA Educational Framework (cf. [Stumpfegger et.al. 2007]) Low level services (changing joint angles, open/close gripper) Transformation services (direct/inverse kinematics) Motion planning services Controlled by the CellRobot service: Take Process Drop The autonomous carts Based on the differential drive with two wheels Controlled by the CellCart service: Drive May 2008 Hoffmann et.al., University of Augsburg, Germany 18
Implementation Adding the self x properties to the production cell 1) Extend the components of the traditional system to gain more degrees of freedom (cf. CellRobot service with multiple tools) 2) Wrap the components in self X services 3) Introduce observer/controller component for the dynamic reconfiguration May 2008 Hoffmann et.al., University of Augsburg, Germany 19
Implementation Implementation ti concentrates t on self healing: li What happens when a tool is breaking down? How to test & demonstrate the self healing abilities of the adaptive production cell? HTTP interface Control GUI manually induce fil failures May 2008 Hoffmann et.al., University of Augsburg, Germany 20
The adaptive production cell in action Movie available: http://www.informatik.uni augsburg.de/lehrstuehle/swt/se/projects/organic_computing/ augsburg computing/ May 2008 Hoffmann et.al., University of Augsburg, Germany 21
Problems & Open Issues Real time behavior bh Underlying problem Loosely coupled ldservices Asynchronous calls No global timing/no control loop Effects No real time constraints No guaranteed response time Inaccurate synchronization of robot operations Sensor input/tool feedback May 2008 Hoffmann et.al., University of Augsburg, Germany 22
Problems & Open Issues Computing power Limitation for the accuracy of the physics engine Limited number of robots & carts Accuracy of the simulation Motion planning in simulation vs. real life Status of workpiece is indicated by color Support of real hardware The production cell onlyruns inthe simulationenvironment May 2008 Hoffmann et.al., University of Augsburg, Germany 23
Future Work Implementing other self X properties Self configuration Self adapting Implementinga framework for controlling industrial robots in Microsoft Robotics Studio Enabling synchronization/coordination of components Introducing a timing/control loop simulation of Slowing downthe physics engine softreal time Introducing additional robots (e.g. KUKA KR6) May 2008 Hoffmann et.al., University of Augsburg, Germany 24
Conclusion Adaptive production systems withself self X properties are more flexible and fault tolerant compared to traditional systems Proofof of concept in Microsoft Robotics Studio Drawbacks: Only self healing until now Asynchronous operations complicate the precise synchronization of components (e.g. robots, carts) No real time constraints Next steps: Implementing other self X properties Implementing a framework for controlling industrial robots in Microsoft Robotics Studio May 2008 Hoffmann et.al., University of Augsburg, Germany 25
Thank you for your attention! Questions? http://www.robotics campus.de p// p May 2008 Hoffmann et.al., University of Augsburg, Germany 26
References [Seebach et.al. 2007] H. Seebach, F. Ortmeier, and W. Reif, Design and Construction of Organic Computing Systems in Proceedings of the IEEE Congress on Evolutionary Computation 2007. IEEE Computer Society Press, 2007, accepted for publication. [Stumpfegger et.al. 2007] T. Stumpfegger, A. Tremmel, C. Tarragona, and M. Haag, A Virtual Robot Control Using a Service Based Architecture and a Physics Based Simulation Environment in Proceedings of the Second International Workshop on Software Development and Integration in Robotics, Rome, 2007 May 2008 Hoffmann et.al., University of Augsburg, Germany 27