Scribd
Upload
19 views

2014 Nonlinear Control

Fully-actuated robot (n joints, n actuators) Dynamic model ‣ M is the positive definite, n by n inertia matrix ‣ C is the n-dimensional vector of Coriolis and centripetal forces ‣ N is the n-dimensional vector of gravitational forces ‣ τ is the n-dimensional vector of actuator forces and torques ‣ is skew symmetric ‣ Two general schemes ‣ Point to point control ‣ Trajectory control

Uploaded by

sandeep2506
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
19 views

2014 Nonlinear Control

Fully-actuated robot (n joints, n actuators) Dynamic model ‣ M is the positive definite, n by n inertia matrix ‣ C is the n-dimensional vector of Coriolis and centripetal forces ‣ N is the n-dimensional vector of gravitational forces ‣ τ is the n-dimensional vector of actuator forces and torques ‣ is skew symmetric ‣ Two general schemes ‣ Point to point control ‣ Trajectory control

Uploaded by

sandeep2506
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

Control of Affine Systems

State x
Input u
State equations

Output

Lie Derivatives
Want
or

Need derivative of the output function

Lie Derivatives

State x

Affine, Single Input Single


Output

Input u
State equations

Output

Relative degree, r

The index of the first nonzero term in the sequence

r=k+1
3

Single degree of freedom arm


1
ml q + mgl sin q =
2

0
x2
x =
+ 1 u
g
sin(x
)
1
l
ml2
2

f (x)

l
q
m

x1
q
x=
=
x2
q

h = x1

g(x)

Lg h = 0

Lf h = x2

1
Lg Lf h =
ml2

L2f h

g
sin x1
l

Affine, SISO
r=1

Linear control,
model independent
feed forward
feedback

r=2

r=3

General form of control law


5

Single degree of freedom arm


1
ml q + mgl sin q =
2

0
x2
x =
+ 1 u
g
sin(x
)
1
l
ml2
2

f (x)

1
Lg Lf h =
ml2

q
m

x1
q
x=
=
x2
q

h = x1

g(x)
L2f h

g
sin x1
l

Fully-actuated robot (n joints, n actuators)


M (q)
q + C(q, q)
q + N (q) =
Dynamic model

M is the positive definite, n by n inertia matrix



C is the n-dimensional vector of Coriolis and centripetal forces

N is the n-dimensional vector of gravitational forces

is the n-dimensional vector of actuator forces and torques

is skew symmetric

Two general schemes

Point to point control


Trajectory control

Fully-actuated robot (continued)


M (q)
q + C(q, q)
q + N (q) =

x1
q
x=
=
x2
q

x =

u = 2 Rn

x2
0
+
M (x1 ) 1 (N (x1 ) + C(x1 , x2 )x2 )
M (x1 )

y = qdes

q 2 Rn

Fully-actuated robot (continued)

Relative degree is 2

u = (Lg Lf h)

Lf Lf h + ydes + k1 (y des

y)
+ k2 (y des

Control law
= M (q) ydes + k1 (y des

y)
+ k2 (y des

y) + C(q, q)
q + N (q)

y)

Visual Servoing

10

Position-Based Visual Servoing


Desired
Cartesian
position/
orientation

Actual Cartesian
position/orientation

Image Based Visual Servoing


Desired position
of feature(s) in
image

Actual feature position

11

Image-Based Visual Servoing


Current view

Desired view

8 outputs, 6 inputs

12

Image plane motion

13

Image plane motion

14

Image Based Visual Servoing

Image-Based Visual Servoing for 3 Features

"

v =

f
u Z

f
v Z

Z
v

u u
v

f
2
f +2v v
2
v f

g(x)
Relative degree is 1

f 2 +2u u
2
u f
v u
v

vx
7
#6
6 vy 7
7
v 6
6 vz 7
6 7
u
6!x 7
4 !y 5
!z

input

u 2 R6

16

Kinematic planar cart


State equations, inputs

Outputs

x + L cos
y = h(x) =
y + L sin

Relative degree is 1
17

Planar Quadrotor

18

Planar Quadrotor

y
y = h(x) =
z
Repeated differentiation of h(x) does not yield explicit dependence on u
Must extend state with higher order derivatives of input
New
extended
state

x
= y
New
input

u
1
u
1
u
=
=
u
2
u2

u1

u 1

T
19

Planar Quadrotor

y
y = h(x) =
z
2

6
6
6
6
6
x
= 6
6 u1
6m
6
6
4

2
y
0
7 60
z
7 6
7 60

7 6
u1
7 60
sin
7+6
m
60
cos
g7
7 6
7 60
0
7 6
5 40
u 1
1
0

f(
x)

Verify

0
0
0
0
0

7
7
7
7
7 u
7 1
7 u
2
7
1 7
Ixx 7
0 5
0

g(
x)

Lg L3fh is full rank

20

Dynamic State Feedback

v=Kz
21

Differential Flatness
All state variables and the inputs can be written as smooth
functions of flat outputs and their derivatives

diffeomorphism

22

You might also like