Course content

The course gives and introduction to methods for adaptation and learning in control of dynamic systems with uncertain parameters. On-line parameter estimation: On-line parameter estimation: gradient methods and least squares methods in continuous and discrete time. Parameter estimation with projection. The Kalman filter as parameter estimator. Dynamic regressor extention and mixing (DREM). Finite time parameter estimation. Extremum seeking methods. Direct and indirect adaptive control: pole placement control (PPC), adaptive pole placement control (APPC), model reference control (MRC), model reference adaptive control (MRAC), adaptive backstepping with tuning functions. Machine learning methods: short introduction to reinforcement learning control.

Learning outcome

Knowledge: The student should posess: - detailed knowledge of on-line parameter estimation and the development and properties of the various methods. - detailed knowledge of adaptive and learning control systems and their development and properties. - detailed knowledge of methods and tools for stability analysis of adaptive and learning systems. Skills: The student should independently be able to: - apply methods for on-line parameter estimation. - develop adaptive and learning control systems. - analyze a problem and select an appropriate method. - analyze existing methods with respect to stability. General skills: The student should be able to: - communicate technical issues with both experts and laymen. - communicate independent work in writing. - contribute to creative processes.

Learning methods and activities

Lectures and compulsory homework assignments.

Compulsory assignments

• Exercises

Recommended previous knowledge

Courses TTK4105 Control Systems and TTK4115 Linear System Theory.

Course materials

To be announced.