Module Number: EI78020
Duration: 1 Semester
Recurrence: Winter Semester
Number of ECTS: 6
Professor in charge: Daniel Müller-Gritschneder
The module assessment contains two parts: An evaluation of the project work (contributing 50% to the final grade) and an oral exam at the end of the semester (contributing 50%).
A) The project work is designed to evaluate the programming and analytical skills of the students in the context of implementing control algorithms on embedded platforms.. In addition, it will also test the ability of the students to understand and use academic/industrial tools like optimization problem solvers, embedded systems design and timing analysis tools and tools for modeling and specifying control algorithms like MATLAB/Simulink. Furthermore, it will also examine students’ understanding of how to systematically implement control algorithms on distributed embedded platforms, which will be taught as a part of this lab through various hands-on exercises. a state-of-the-art control/architecture co-design methodology for embedded control systems (which will be explained to them). Therefore, the project work will consists of several programming and implementation tasks. In case of programming tasks, the code will be tested for correctness of functionality and efficiency. Solution to the implementation tasks will be examined via demonstrations and conversations.
B) In the oral exam, the students need to demonstrate their understanding of the concepts of feedback control theory, embedded systems, control/architecture co-design and optimization theory.
These prerequisites are desirable. Not all need to be known.
1. Basics of control theory (State feedback control, Continuous-time and discrete-time mathematical model, State-space representation)
2. Basics of Embedded Systems (Time-triggered and Event-triggered scheduling)
After successfull completion of the module, students gained basic knowledge of embedded control systems. First, they are able to analyze an embedded platform architecture commonly used in automotive industry and derive implementation constraints from the same. Second, they know the theory of feedback control and how a feedback controller is designed considering constraints from the implementation platform as well as from the control side. Third, they know how to implement a controller on an embedded platform. Finally, they are able to understand and implement a state-of-the-art control/architecture co-design methodology to design and implement embedded control systems. Furthermore, they are familiar with some academic/industrial tools used in this domain. In addition, they are able to formulate and solve linear and non-linear optimization problems and will also use commercial optimization tools.
Embedded control systems are commonly found in various application domains like automotive, avionics, industry automation, etc. Often these systems are safety-critical and must guarantee certain level of safety. However, traditionally controllers are designed separately in MATLAB/Simulink using closed-loop simulation of plants and controllers. Subsequently, they are implemented using some platform design tools where the control algorithm is considered as a black box. Nevertheless, there is a strong interplay between the control algorithms and the platform architecture which if not considered can jeopardize the safety of the system. In particular, traditional control theory courses focus on the basics of designing control algorithms and analyzing them. However, stability and other control performance guarantees at the design stage do not carry over to an implementation on a distributed embedded platform. Here, mostly ad hoc techniques and a lot of testing, debugging and iterative refinements are used to obtain a satisfactory implementation of a control algorithm. In this laboratory course we will teach the students how to design safe embedded control systems using a state-of-the-art control/architecture co-design methodology. In other words, the focus will be to teach techniques for systematically implementing control algorithms in a hands-on manner.
Towards this we will teach the following topics:
1. Feedback control theory a. Continuous and discrete-time mathematical models b. State-space representation c. Stability of closed-loop systems d. Control performance metrics e. Pole-placement technique for controller design f. Evolutionary algorithm for optimal control design
2. Embedded systems theory a. Time-triggered and event-triggered scheduling b. Schedule parameters c. Architectural and scheduling constraints d. Mixed-Integer Linear Programming (MILP) formulation of schedule synthesis problem.
3. Control/Scheduling Co-design a. Optimization objectives b. Interplay between control and scheduling parameters c. MILP formulation of the optimization problem d. Solution using a nested two-layer optimization technique
Furthermore, students are required to implement and simulate their design results on academic/industrial tools.
1) The concepts of feedback control theory and embedded systems will be introduced via lectures.
2) The state-of-the-art control/architecture methodology will be explained by the supervisor which the students need to implement.
3) A short overview of the software tools will be provided by the supervisor.
4) Students must go through software manuals for more details of the tools.
5) Students must go through the laboratory manual for the details of each task to be executed.
6) Students will also be provided research papers which may be useful for better understanding.
Lecture slides, software manuals and laboratory manual. Embedded systems design tool, MATLAB toolboxes, Simulink libraries
1) Embedded Control Systems (EI7262) Lecture Slides and Tutorials.