Control of Autonomous Systems

Welcome!

Jerome Jouffroy, Professor

jerome@[Link]
Practicalities of the course
- Every Monday or Wednesday

- Course in parallel in Odense and Sønderborg

- About 10 or 11 lectures

- Slides and Lecture notes will be posted on Itslearning for

each lecture

- Written exam with theoretical and programming questions

Examination

- 10 questions
- 2 points each question = 20 max points for the total
- 50%=10 points to pass (grade 02), 20 points to ace (grade 12)
- theoretical (pen and paper) questions and programming questions
(ie implement this or that in Matlab/Simulink)
- Theorecical questions: be precise and justify your answer
- Programming questions: it has to work and give the expected
behavior
- Programming questions: I need to be able to open the program on
my computer (also: do not program like a “bozo” J!)
Syllabus
- Dynamical systems
- State-space representations
- IO representation and PID control
- Stability and state-feedback control
- Tracking and dynamic state-feedback
- MPC and feedback linearization
- State estimation and observers
- Output feedback and ADRC
- Parameter estimation
- Controllability and feedforward control
- (Differential flatness and trajectory planning)
Lectures and teaching philosophy
- Learning by doing is the essence of this course

- Each lecture will have a 50 to 60-minute presentation to introduce a

concept/technique, followed by a computer session where we practise
the technique

- All exercises should be finished for the next time

- Attitude vs skills… (be professional, prepared, etc.)

- There will be an assignment midway through the semester

 Lecture 1

Dynamical systems and block diagrams

Jerome Jouffroy, Professor
jerome@[Link]
Signal and system representations
In the beginning there was…

Plant

the dynamical system Σ

where: - u(t) signal having effect on the behavior of system Σ,
ie u(t) is called (control) input
- y(t) signal representing a selected part of the dynamical
system/plant dynamics,
ie y(t) is called output
A few systems (1/7)
Bank account with interest rates
Maybe one of the simplest
dynamical there exist…
At year t=0, a bank account with 3% interest rate is opened with initial deposit
(say 10000DKK).
The evolution of the amount y at year t, ie y(t), is given by

where
recurrence relation initial condition
The owner of the account adding/withdrawing some
money can be represented by an input variable:

Remark: 1 input, 1 output, linear system, discrete-time

A few systems (2/7)
Mass-Spring-Damper system

F=u

Dynamical system represented by an Ordinary Differential

Equation (ODE):

with the initial conditions initial position of the mass

initial velocity of the mass
Remark: 1 input, 1 output, linear system, continuous-time
A few systems (3/7)
Pharmacokinetics

Represented by

with C1, C2 concentration of drug in compartment i, u is

colume flow rate of drug intake, V1, V2 volumes of
compartments i.
monitored output
Remark: 2 coupled dyn. syst., 1 input, 1 output, linear
A few systems (4/7)
Pendulum

Represented by
with
where we measure Remark: nonlinear system, no input!

add damping:
add input/torque:
with
A few systems (5/7)
Lorenz system

Ever heard about the butterfly

effect? That is the equation:

where p is a constant
Remark: 3 coupled nonlinear dyn. syst., no input
A few systems (6/7)
2-wheeled robot

u1: longitudinal velocity of robot,

u2: rotational velocity This gives

We want to monitor all 3 dynamical variables, ie we have the outputs

Remark: nonlinear dyn. syst., 2 inputs, 3 outputs

A few systems (7/7)
Car-like robot

Very similar to the 2-wheeled robot:

with
Block diagrams: the basic blocks
Graphical representation for dynamical systems.
Use a library of blocks…

summator (sum) gain

differentiator integrator
…and connect them to
each other to obtain a
block diagram.
multiplier
Block diagram on a simple example (1/2)
Objective: make a block diagram of the system

with

BUT: - Initial condition not

represented
- Differential operator
very sensitive to
noise!

Another way?
Block diagram on a simple example (2/2)
Use a nice trick/property:

which gives

So that gives the block diagram

OR
Parenthesis: story behind the integrator trick/method

William Thomson
aka Lord Kelvin

“differential analyser”
 F=u
Exercise (pen and paper/tablet)

Make a block diagram (with initial

conditions) of
A word on discrete-time systems
Working with discrete-time systems is very similar to continuous-time ones.
Instead of an integrator, use
a unit delay block:

unit delay
As an example, our bank account
system represented by

gives
Exercises with MATLAB/Simulink

- Questions for the exercise of this lecture are posted on

Itslearning

- Have you downloaded MATLAB/Simulink? J