1 of 49

Aerial Robotics

Control Theory

C. Papachristos

Robotic Workers (RoboWork) Lab

University of Nevada, Reno

CS-491/691

2 of 49

Signal Processing Fundamentals

  • Fourier Transform

  • A mapping between representations in different domains:
    • Time Domain

    • “Frequency” Domain

  • Sinusoids are the only functions (Note: together with exponentials) that retain their “shape” when differentiated
    • Physical Laws work in Differential Equations
    • We can associate the transformation of a signal fed through a dynamical system as an operation/transformation on its Frequency Domain features (Amplitude & Phase)

CS491/691 C. Papachristos

 

 

Time (s)

Frequency - 2πω (rad/s)

 

Amplitude

Frequency - 2πω (rad/s)

Phase

Dynamical System

Differential Equations

Original�Amplitude �& Phase

Sinusoid

Input

Modified�Amplitude �& Phase

Sinusoid

Output

Note: Indicative images of different domains, but for the Fourier Series

3 of 49

Signal Processing Fundamentals

 

CS491/691 C. Papachristos

 

 

 

 

 

 

 

 

 

 

 

4 of 49

Signal Processing Fundamentals

 

CS491/691 C. Papachristos

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5 of 49

Signal Processing Fundamentals

 

CS491/691 C. Papachristos

 

 

 

 

 

 

For

ω=3

 

 

For

ω=5

By Thenub314 - Own work, CC BY-SA 3.0

 

 

Fourier Transform Magnitude

 

6 of 49

Linear Time Invariant Dynamical Systems

  • Linear Time Invariant (LTI) dynamical systems are characterized by:

    • Homogeneity (Linearity):

    • Superposition (Linearity):

    • Time Invariance:

  • Important class of systems because “We can solve them”, i.e. we have developed a set of mathematical tools to systematically solve them
  • Also, we can approximate many physical processes with LTI dynamical systems

CS491/691 C. Papachristos

 

L T I

Dynamical System

 

 

 

 

7 of 49

Modeling of LTI Dynamical Systems

 

Time Domain based representation

Complex Domain based representation

 

 

CS491/691 C. Papachristos

 

 

 

 

 

 

 

 

 

 

 

 

 

8 of 49

Modeling of LTI Dynamical Systems

 

Time Domain based representation

Complex Domain based representation

 

 

CS491/691 C. Papachristos

E.g.,:

 

 

 

 

 

 

 

9 of 49

Modeling of LTI Dynamical Systems

 

 

 

 

 

 

 

 

 

Note 2: Used in Systems Analysis

  • Transforms linear Differential Equations into Algebraic ones
  • Transforms Convolution into Multiplication

 

CS491/691 C. Papachristos

10 of 49

Modeling of LTI Dynamical Systems

 

Time Domain based representation

Complex Domain based representation

 

 

CS491/691 C. Papachristos

 

 

 

11 of 49

Modeling of LTI Dynamical Systems

 

 

 

CS491/691 C. Papachristos

L T I

Dynamical System

Note: The Transfer Function is the Laplace Transform of the LTI’s Impulse Response

 

 

 

 

 

 

12 of 49

Modeling of LTI Dynamical Systems

 

 

 

 

 

CS491/691 C. Papachristos

 

 

13 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

14 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

15 of 49

Analysis of LTI Dynamical Systems

 

C. Papachristos

 

Gain Origin z/p Real z/p Complex z/p Transp. Delay

On the whiteboard:

 

16 of 49

Analysis of LTI Dynamical Systems

 

 

 

 

 

CS491/691 C. Papachristos

17 of 49

Analysis of LTI Dynamical Systems

 

 

 

 

 

CS491/691 C. Papachristos

Step Response

18 of 49

Analysis of LTI Dynamical Systems

2nd Order Systems Response Analysis

Second order system representation:

In transfer function form:

DC-gain :

Damping ratio:

Natural frequency:

Poles:

 

 

 

 

 

 

 

 

CS491/691 C. Papachristos

19 of 49

Analysis of LTI Dynamical Systems

 

 

 

 

 

 

CS491/691 C. Papachristos

Note: A critically damped system reaches the Steady-State Response in the minimum time possible

20 of 49

Analysis of LTI Dynamical Systems

2nd Order Systems Response Analysis

Pole Maps on the Complex Plane:

 

 

 

 

Poles are Complex numbers

Poles have only Real components

Poles have only Imaginary component

Complex Root-Pair

Locus:

CS491/691 C. Papachristos

constant “damping ratio”

lines

21 of 49

Analysis of LTI Dynamical Systems

2nd Order Systems Response Analysis

Time-Domain Responses:

 

 

 

 

Some overshoot & oscillation exists

No overshoot (or oscillation)

Fixed-amplitude oscillation

Complex Root-Pair

Locus:

CS491/691 C. Papachristos

Step Response

Step Response

Step Response

constant “damping ratio”

lines

22 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

 

23 of 49

Analysis of LTI Dynamical Systems

 

 

CS491/691 C. Papachristos

24 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

 

25 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

 

26 of 49

Analysis of LTI Dynamical Systems

 

Critical

Stability

CS491/691 C. Papachristos

 

27 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

 

No oscillations

oscillations

oscillations

Stable

28 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

 

29 of 49

Analysis of LTI Dynamical Systems

 

C. Papachristos

Note: A right-half-plane Zero’s effect on a (step) response is an initial “dip” in the wrong direction (harder to control!)

A Non-Minimum-Phase system:

The Minimum Phase system:

Another Non-Minimum Phase system:

 

 

 

30 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

31 of 49

Analysis of LTI Dynamical Systems

 

 

 

 

 

C. Papachristos

32 of 49

Analysis of LTI Dynamical Systems

 

C. Papachristos

 

33 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

Note: Upper-half of Nyquist plot is a mirror of lower-half (corresponds to response for negative frequencies)

 

34 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

 

 

Unstable Zeros

(or closed-loop Unstable Poles)

 

 

(only) Unstable open-loop Poles

35 of 49

Analysis of LTI Dynamical Systems

Stability Margins

  • Bode-based Gain & Phase Margins inform us of tolerance when only one of either the Gain/Phase changes

  • Bode plot not useful for Unstable / Marginally Unstable systems

  • Discuss how to find margins on Nyquist

  • Discuss Disc Margins

  • Low-pass / High-pass / Band-pass / Notch filters?

CS491/691 C. Papachristos

36 of 49

Analysis of LTI Dynamical Systems

 

 

 

CS491/691 C. Papachristos

 

 

 

37 of 49

Analysis of LTI Dynamical Systems

 

CS491/691 C. Papachristos

 

 

 

 

38 of 49

Analysis of LTI Dynamical Systems

 

C. Papachristos

 

 

 

 

39 of 49

State-Space Modeling of LTI Dynamical Systems

 

 

 

 

CS491/691 C. Papachristos

40 of 49

State-Space Modeling of LTI Dynamical Systems

 

 

 

CS491/691 C. Papachristos

41 of 49

State-Space Modeling of LTI Dynamical Systems

 

 

 

 

 

 

 

CS491/691 C. Papachristos

42 of 49

State-Space Modeling of LTI Dynamical Systems

Example

SISO system Ordinary Differential Equation (2nd order):

In State-Space form:

(In State-Space form: )

 

 

 

 

 

 

 

 

 

 

CS491/691 C. Papachristos

43 of 49

State-Space Modeling of LTI Dynamical Systems

 

 

C. Papachristos

Eigenvalues Equation

 

 

 

 

44 of 49

State-Space Analysis of LTI Dynamical Systems

 

 

 

Continuous

Time:

C. Papachristos

 

45 of 49

 

State-Space Analysis of LTI Dynamical Systems

 

 

 

 

 

Continuous

Time:

Discrete

Time:

 

CS491/691 C. Papachristos

 

46 of 49

 

State-Space Analysis of LTI Dynamical Systems

 

 

Continuous

Time:

Discrete

Time:

 

 

CS491/691 C. Papachristos

47 of 49

State-Space Analysis of LTI Dynamical Systems

 

 

 

 

 

Continuous

Time:

Discrete

Time:

 

CS491/691 C. Papachristos

48 of 49

State-Space Analysis of LTI Dynamical Systems

 

 

 

 

Continuous

Time:

Discrete

Time:

 

CS491/691 C. Papachristos

49 of 49

Time for Questions !

CS-491/691

CS491/691 C. Papachristos