| Conventionally, fixed parameter controllers are used for (motion) control. Such controllers are widely applied because of the ease of design, stability guarantees and satisfactory performance. However, performance can often be improved significantly by updating the controller using information on the disturbance or system dynamics learned from data. Such learning and adaptive controllers are considered in this course. |
In the first part of the course, the underlying principles of the learning and adaptive control techniques are introduced and basic implementations and typical applications are presented. Participants also implement basic versions of the controllers in simulation. In the second part of the course, the participants study and evaluate recent developments in fundamentals, algorithms and/or applications of learning and adaptive control techniques in literature. This can be related to a project or challenge of the participants’ interest. In the second part also some guest lectures are scheduled to show the developments and use of the learning and adaptive control in research and industry.
Why this course: This course considers controllers that learn disturbances, particularly disturbance observer based control (DOBC) for disturbances with known dynamic behaviour and iterative learning control (ILC) for repeating disturbances. Furthermore, the course considers controllers that learn the system dynamics, particularly adaptive feedforward control (AFC) learning the dynamic response to a known reference or disturbance and model reference adaptive control (MRAC) learning the controller parameters to match a prescribed closed-loop response. Finally, the course considers adaptive control of uncertain Euler-Lagrange systems (ACUELS) to deal with the parametric uncertainties in the context of Lyapunov-based design and analysis techniques.
Within this course, the following knowledge and skills will be acquired:
- Explanation of basic concepts, assumptions, algorithms and typical applications of learning and adaptive control methods, including disturbance observer based control (DOBC), iterative learning control (ILC), adaptive feedforward control (AFC), model reference adaptive control (MRAC) and adaptive control of uncertain Euler-Lagrange systems (ACUELS);
- Implementation of the learning and adaptive control methods and simulate their response;
- Summary and evaluation of a recent development or application of learning and adaptive control methods from a literature study.
Course highlights:
- Performance beyond conventional control;
- Discussion and analysis of basic learning and adaptative control methods;
- Guided in-depth study on topic of interest.
For whom: Professionals with knowledge of calculus, linear algebra, linear systems, basic dynamical modelling (of mechanical systems), state-space representations, digital control, optimal control and control of MIMO systems is required. If needed, this control knowledge can be obtained through the Control System Design for Robotics, given at the UT as well.
From whom: prof.dr.ir. W.B.J. Hakvoort, dr. H. Köroğlu, dr. J. Dasdemir
Practical information: This is a regular master course, in which students as well as professionals can participate. The classes can be streamed; presence is required for feedback on the assignments. For the final assignment, the topic can be chosen individually; it is done in self-study.
Location: University of Twente, Enschede, NL
Duration: The course is scheduled annually from February till April. It requires 140 hours of study load.
Costs: € 2067,15
More information:
Content of the course: prof.dr.ir. W.B.J. Hakvoort, w.b.j.hakvoort@utwente.nl
Registration: Registration form | Faculty of Engineering Technology (ET)