Discover how controllers learn from data and adapt to disturbances and changing systems to improve performance in motion control and robotics.
Learning and Adaptive Control explores how controllers can improve their performance by learning from data about disturbances or changing system dynamics. Conventional controllers use fixed parameters and are widely applied because they are relatively straightforward to design and offer stability guarantees. However, fixed controllers cannot always maintain optimal performance when systems or operating conditions change.
In this course, you study control methods that learn disturbances or adapt to uncertain system dynamics. Topics include disturbance observer-based control, iterative learning control, adaptive feedforward control, model reference adaptive control and adaptive control of uncertain Euler-Lagrange systems. You learn the underlying concepts, assumptions and algorithms, implement basic versions of these controllers and simulate their responses. The course helps you assess which learning or adaptive method is suitable for a specific control problem and how it can improve performance beyond conventional control.
Modern motion systems and robotic applications increasingly operate under changing conditions, uncertain system parameters and recurring disturbances. Fixed-parameter controllers may provide satisfactory performance, but they do not use new information to improve their response.
Learning and adaptive control methods use measured data to update controller behaviour. They can learn repeating disturbances, estimate system dynamics or adjust controller parameters to achieve a desired response. These methods are increasingly relevant for professionals working with precision systems, robotics, mechatronics and advanced motion control where consistent performance and adaptability are important.
This is a regular master’s course in which both students and professionals can participate. In the first part of the course, the underlying principles of learning and adaptive control are introduced. Participants study basic implementations, typical applications and the assumptions behind different control methods. Weekly assignments provide experience in implementing the methods and simulating their responses.
In the second part, participants study and evaluate a recent development or application from scientific literature. Professionals can select a paper or topic connected to a project or challenge from their own organisation. Guest lectures provide examples of recent developments and applications in research and industry.
The course can be followed partly online by professional participants. Attendance is required for assignment feedback sessions. Where possible, assignments and assessment can be adjusted to reflect the professional participant’s learning goals. Possibilites can be disscussed with the teaching staff.
Details and assessment:
This course is intended for professionals who work with motion control, robotics, mechatronics or advanced dynamical systems and want to improve system performance through learning and adaptation. Typical profiles include, but are not limited to, control engineers, robotics engineers, mechatronics engineers, R&D engineers, system architects, motion control specialists and technical researchers.
The course requires prior knowledge of calculus, linear algebra, linear systems, dynamical modelling of mechanical systems, state-space representations, digital control, optimal control and control of MIMO systems. This background can, for example, be obtained through the University of Twente course Control System Design for Robotics.
If you successfully complete the course, you will receive a microcredential. This microcredential is a digital record confirming that you have passed the assessment components of the regular master’s course and achieved the corresponding learning outcomes.
You may also participate without completing the assessments. If you choose this option, or do not pass all assessment components, you will receive a digital certificate of participation.
Start date: February (Q3)
Total duration: February to April (excluding resits) (140 hour study load)
Number of teaching days/sessions: Lectures, tutorials, weekly assignments, guest lectures, feedback sessions and an oral examination. The exact number of sessions is to be confirmed.
Total programme fee: €2067.15 (5 ECTS × €413.43 per ECTS)
Programme type: Master course
Location: University of Twente, Enschede, NL
Language of instruction: English
Admission requirements: 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. The course Control System Design for Robotics, or equivalent prior education, provides the expected control engineering background.
Study materials:
Teachers: