Thomas Holzhüter
Research Projects

Optimal Control of Nonlinear Systems

The nonlinear optimal control problem can generally be solved by integrating the Hamilton-Jacobi partial differential equation.
We work on methods to numerically calculate the optimal controller for the infinite horizon, i.e. time invariant problem.
A first approach was via backward integration of the Euler-Lagrange ordinary differential equations, see:

Th. Holzhüter (2004). Optimal Regulator for the Inverted Pendulum via Euler-Lagrange Backward Integration   (ps)   (pdf)
Automatica, Vol. 40  No. 9 (2004),  pp. 1613-1620.

Th. Holzhüter and Th. Klinker (2007). Method to solve the nonlinear infinite horizon optimal control problem
with application to the track control of a mobile robot.  
International Journal of Bifurcation and Chaos Vol. 17, No. 10, pp. 3607-3611.

Currently we work on a numerical procedure using grid methods for solving static Hamilton-Jacobi equation. Details will be available later.

Mobile Robots  

We use AmigoBot robots controlled from the MATLAB/SIMULINK environment.
This environment is also used in course on RealTimeSystems in
the Master Course on Information Engineering at HAW Hamburg.

Master Projects
        Fang, Li (2004).  Camera based track control for a mobile robot.   pdf
        Wu, Kai (2004).  Realtime control of a mobile robot using MATLAB. pdf
        Ye, Guanghua & Wang, Quingyan (2005). Observation camera based navigation system for mobile robots  pdf(zip)
        Chung,  Joseph Man-Kit (2006).  Formation Control for Multiple Mobile Robots   pdf

Recently we started using the ROBOTINO mobile robot  produced by Festo.
Currently we work on driving the Robot remotely from Matlab/Simulink.
Some  information for in-house use  is available here,  further results will be published later.

Rapid Prototyping  and Automatic Code Generation via Matlab/Simulink

In connection with an industrial partner and as a focus for the coming Master Course in Automatic Control
at the Department of Electrical and Information Engineering we are developing
a sample implementation, using the Simulink xPC real time target. Results will be presented later.