IFAC World Congress

Workshop July 12, 2020

Berlin, Germany

 

 

 

Iterative Learning Control: Theoretical Foundations and Design Frameworks

 

Tom Oomen* Bing Chu** Kira Barton***

 

*Eindhoven University of Technology, Eindhoven, The Netherlands
(e-mail: t.a.e.oomen@tue.nl).
** University of Southampton, United Kingdom
 (e-mail: b.chu@soton.ac.uk)
*** University of Michigan, Ann Arbor, Michigan, USA
(e-mail: bartonkl@umich.edu)

 

 

1. WORKSHOP DESCRIPTION

Humans learn every day to improve their skills. In sharp contrast, traditional control systems, including feedforward and feedback, do not improve their performance, while the amount of available data is drastically improving in recent years. This workshop aims to provide the basics for and ongoing research on data-driven intelligent control systems by learning from data.

These data-driven methods include Iterative Learning control (ILC) and repetitive control (RC), which are high performance tracking control design methods for systems operating in a periodic or repetitive manner. To achieve this, they both adapt the control effort based on information collected from previous trials (periods). Compared to conventional control design approaches, ILC and RC potentially lead to significantly better performance even without accurate system model information. Originating from robotic research, ILC and RC have attracted intensive research effort and have proven to be extremely successful in achieving attractive system performance in a wide range of application domains, including manufacturing processes, mechanical testing equipment, chemical batch processes and next-generation health care.

On a fundamental level, such approaches, including iterative learning control (ILC), lead to a 2D control system, comprising both feedback and feedforward elements, and their stability/convergence is analyzed in detail using control theoretic techniques. A complete design framework is provided, covering both approaches from loop-shaping (control engineering) and optimization (optimal control). Furthermore, multivariable systems, reset vs. reset-free operation (repetitive control), flexible learning via basis functions (learning feedforward), the role of modeling and data-driven learning (IIC algorithms) are central elements in recent developments.

After more than 30 years, ILC and RC have progressed considerably in both theoretical research and its practical application. This workshop, together with an open invited track within the main conference, aims to provide an overview of the latest advances in ILC and RC and to create a forum for high quality discussion of both theoretical and practical perspectives. In particular, the workshop aims to:

·       Bring together results representing the dominant analysis and design paradigms, including frequency domain design, norm-optimal design, internal model design.

·       Address fundamental aspects in the design of ILC algorithms, including noncausality, convergence vs. monotonic convergence, etc.

·       Address new theoretical challenges in ILC and RC, including robustness and flexibility to varying tasks, networked systems, etc.

·       Present new emerging and non-traditional applications.

·       Discuss future challenges and opportunities in ILC and RC.

2. TARGET AUDIENCE

Do you also have a system that has the same error for each task? This workshop enables you to learn how to improve the performance of your control system. By the end of the workshop, you will be able to:

·       identify (traditional and emerging) applications of ILC and RC,

·       design ILC and RC using state-of-the-art methodologies to improve the system’s performance, and

·       develop knowledge of future research challenges and opportunities in ILC and RC.

A basic knowledge of feedforward and feedback control is essential, while working knowledge of linear algebra and basic system theory suffices to get familiar with the main ideas. No prerequisite knowledge in ILC and RC is needed.

3. PRELIMINARY PROGRAM

9.00 - 9.15: Welcome and Introduction (all)

9.15 - 10.00: ILC analysis and design in frequency domain (Oomen)

10.00 - 10.30: Time domain ILC analysis and design using lifted form representations (Oomen)

10.30 - 11.00: Break

11.00 - 12.00 ILC using optimisation based approaches (Chu)

12.00 - 12.30 Experimental demo (Eindhoven-PhD)

12.30 - 14.00 Lunch

14.00 - 14.45 Spatial ILC (Barton)

14.45 - 15.30 ILC for point-to-point tracking tasks (Chu)

15.30 - 16:00 Break

16.00 - 16.30 ILC design using basis functions (Oomen)

16.30 - 17.00 Emerging ILC approaches (Barton)

4. WORKSHOP SPEAKERS

Professor Tom Oomen, Eindhoven University of Technology, The Netherlands

Biography: Tom Oomen received his MSc and PhD degree from the Eindhoven University of Technology, Eindhoven, The Netherlands. He has held long-term visiting positions at KTH, Stockholm, Sweden, and at The University of Newcastle, Australia, in addition to numerous short visits to international research centers. He is a recipient of both the VENI (2013) and VIDI (2017) personal research grants. He is a senior member of the IEEE, and is presently Associate Editor of IFAC Mechatronics and the IEEE Control Systems Letters (L-CSS). He has been Associate Editor on the IEEE Conference Editorial Board, as well as special issue guest editor for IFAC Mechatronics. He regularly organizes special sessions at international conferences, as well as workshops for academia and industry. He is a member the Eindhoven Young Academy of Engineering (EYAE).

Professor Kira Barton, University of Michigan, USA

Biography: Professor Kira Barton received her B.S. degree in Mechanical Engineering from the University of Colorado at Boulder in 2001. Barton continued her education in mechanical engineering at the University of Illinois at Urbana-Champaign and completed her M.S. and Ph.D. degrees in 2006 and 2010, respectively. She held a postdoctoral research position at the University of Illinois from Fall 2010 until Fall 2011, at which point she joined the Mechanical Engineering Department at the University of Michigan at Ann Arbor. Her primary research focus is on precision coordination and motion control for emerging applications, with a specialization in iterative learning control. Barton’s work intersects controls and manufacturing and combines innovative manufacturing processes with enhanced engineering capabilities. The potential impact of this research ranges from building high-resolution DNA sensors for biological applications, to the integration of advanced sensing and control for rehabilitation robotics.

Professor Bing Chu, University of Southampton, UK

Biography: Dr Bing Chu is an associate professor in Electronics and Computer Science at University of Southampton, UK. Before joining University of Southampton in 2012, he was a postdoctoral researcher at University of Oxford (2010-2012). He obtained his BEng degree in Automation and MSc degree in Control Science and Technology from Tsinghua University, Beijing, China in 2004 and 2007, respectively. In 2009, he completed a PhD degree in Automatic Control and Systems Engineering from the University of Sheffield, UK. His current research interests include iterative learning and repetitive control, analysis and control of large scale networked systems, applied optimisation theory, and their applications to robotics, power electronics and next generation healthcare. He has been an associate editor of International Journal of Control, a member of IEEE Control Systems Society Conference Editorial Board, general chair/programme chair of 2019-IEEE-NDs and 2019-IFAC-ALCOS, as well as a member of several IEEE and IFAC technical committees.

REFERENCES

      Oomen, T. (2018). Learning in Machines. Mikroniek, 6:5-11, 2018. Available online: http://www.dct.tue.nl/toomen/files/Oomen2018b.pdf