Introduction to Robotics (CS223A) Homework #6 Solution

(Winter 2007/2008)
1. Consider the 1-DOF system described by the equation of motion, 4 x+20 x+25x =
f.
(a) Find the natural frequency
n
and the natural damping ratio
n
of the
natural (passive) system (f = 0). What type of system is this (oscillatory,
overdamped, etc.) ?
Using Section 7.2.2 from the course reader, we can compare this sytem with m x + b x +
kx = 0 like in Equation 7.9. Thus:

n
=

k
m
=
_
25
4
= 2.5

n
=
b
2

km
=
20
2

25 4
= 1
Since
n
= 1, this system is critically damped.
(b) Design a PD controller that achieves critical damping with a closed-loop
stiness k
CL
= 36. In other words, let f = k
v
xk
p
x, and determine the gains
k
v
and k
p
. Assume that the desired position is x
d
= 0.
The original system is:
4 x + 20 x + 25x = f
with input force f. The controller provides this input, using the formula:
f = k
v
x k
p
x
So, the closed loop equation is:
4 x + 20 x + 25x = k
v
x k
p
x
4 x + (20 + k
v
) x + (25 + k
p
)x = 0
This closed loop system behaves just like the natural dissipative system in Section 7.2.2
of the course reader. So, we rst compare to Equation 7.9:
4 x + (20 + k
v
) x + (25 + k
p
)x = m x + b x + kx = 0
The closed loop stiness is given by k, the coecient of the positional term, so:
k = 25 + k
p
= k
CL
= 36
For the damping requirement, we need to rst gure out how the control gains k
p
and
k
v
aect the damping ratio ; we do this by applying Equation 7.12 to our system:
=
b
2

km
For critical damping:
= 1
so the coecient b of x must satisfy
b = 2

km = 2

36 4 = 24
Based on our closed loop equation, we have:
b = 20 + k
v
so the gains that we need are
k
p
= 11, k
v
= 4
and the PD controller is
f = 4 x 11x
(c) Assume that the friction model changes from linear (20 x) to Coulomb fric-
tion, 30sign( x). Design a control system which uses a non-linear model-based
portion with trajectory following to critically damp the system at all times
and maintain a closed-loop stiness of k
CL
= 36. In other words, let f = f

+
and f

= x
d
k

v
( x x
d
) k

p
(x x
d
). Then, nd f,,,f

,k

p
and k

v
. Note that
f is an m-mass control, and f

is a unit-mass control. Use the denition of

error, e = x x
d
.
The dierential equation for the system is now
4 x + 30sign( x) + 25x = f
In order to linearize it, we apply a force f of the form
f = f

+
where
= 4, = 30sign( x) + 25x
For purposes of control, this makes the system look like the unit-mass system:
x = f

to which we apply the control

f

= x
d
k

v
( x x
d
) k

p
(x x
d
)
Substituting into our unit-mass system yields the equation
e + k

v
e + k

p
e = 0
where e is the position error, e = x x
d
.
Now, we want to choose our gains k

p
and k

v
so that we achieve critical damping and the
desired closed-loop stiness.
To look at the closed loop stiness, we need to consider the controlled system before
factoring out the mass:
4 x + 30sign( x) + 25x = f

+
4 x = f

4 x + f

= 0
4( x x
d
) + 4k

v
( x x
d
) + 4k

p
(x x
d
) = 0
4 e + 4k

v
e + 4k

p
e = 0
The coecient of the e term is the closed loop stiness, so: k

p
= 9.
In order to have critical damping, we need to have = 1. Using Equation 7.12 we see
that the coecient of e must be:
k

v
= 2
_
k

p
= 6
Thus, the control is
f = f

+
= 4
= 30sign( x) + 25x
f

= x
d
6( x x
d
) 9(x x
d
)
(d) Given a disturbance force f
dist
= 4, what is the steady-state ( e = e = 0) error
of the system in part (c)?
We can analyze the error by observing the error in the unit-mass system. With a
disturbance force added, the systems equation of motion becomes
4 x + 30sign( x) + 25x = f + f
dist
To linearize the system, we apply a force of the same form as before:
f + f
dist
= 4f

+ 30sign( x) + 25x + f
dist
= 4
_
f

+
f
dist
4
_
+ 30sign( x) + 25x
This yields a unit-mass system as before, but now it has a disturbance force of f
dist
/4,
so the unit-mass system now looks like
x = f

+
f
dist
4
With the control from before, we get a unit-mass closed-loop system of
e + 6 e + 9e =
f
dist
4
For the steady state, when x = x = 0, we get
9e =
f
dist
4
So, the steady state error is given by
e =
f
dist
4 9
=
4
36
0.111
2. For a certain RR manipulator, the equations of motion are given by
_
4 + c
2
1 + c
2
1 + c
2
1
_ _

2
_
+
_
s
2
(

2
2
+ 2

2
)
s
2

2
1
_
=
_

1

2
_
(a) Assume that joint 2 is locked at some value
2
using brakes and joint 1
is controlled with a PD controller,
1
= 40

1
400(
1

1d
). What is the
minimum and maximum inertia perceived at joint 1 as we vary
2
? What
are the corresponding closed-loop frequencies?
For joint 2 locked (

2
=

2
= 0), the equation of motion for joint 1 is:
(4 + c
2
)

1
=
1
The inertia seen at joint 1 is the coecient of the

1
term, (4 + c
2
). So, this inertia
achieves its maximum and minimum values at
2
= 0 and
2
= 180

:
m
max
= 5, m
min
= 3
The closed-loop equation for joint 1 is
(4 + c
2
)

1
+ 40

1
+ 400(
1

1d
) = 0
To get an expression for closed loop frequency, we compare our closed loop equation
with the generic system of Equation 7.9 (m x + b x + kx = 0).
The closed loop frequency is then given by:
=

k
m
=

400
(4 + c
2
)
So, we have
m = m
max

min
=
20

5
m = m
min

max
=
20

3
(b) Still assuming that joint 2 is locked, at what values of
2
do the minimum
and maximum damping ratios occur? What are the minimum and maximum
damping ratios?
To get an expression for damping ratio, we once again compare our closed loop equation
with the generic system of Equation 7.9 (m x+b x+kx = 0). In this case, using Equation
7.12 the closed-loop damping ratio is given by:
=
b
2

km
=
40
2

400m
=
1

m
=
1

4 + c
2
So, the minimum and maximum values of occur at
2
= 0 and
2
= 180

min
=
1

m
max
=
1

max
=
1

m
min
=
1

3
(c) Now assume that both joints are free to move, and that this system is
controlled by a partitioned PD controller, =

+ . Design a partitioned,
trajectory-following controller (one that tracks a desired position, velocity
and acceleration) which will provide a closed-loop frequency of 10 rad/sec
on joint 1 and 20 rad/sec on joint 2 and be critically damped over the entire
workspace. That is, let

_
k

v
1
0
0 k

v
2
_
(

d
)
_
k

p
1
0
0 k

p
2
_
(
d
),
then nd the matrices and and the vector , along with the necessary
gains k

v
i
and k

p
i
.
The equations of motion are of the form
M()

+ V (

, ) =
to which we apply a vector of torques of the form
=

+
To make this look like a unit-mass system, we let
= M(), = V (

, )
which gives the unit-mass system

To this system, we apply the control

_
k

v
1
0
0 k

v
2
_
(

d
)
_
k

p
1
0
0 k

p
2
_
(
d
),
This yields two closed-loop equations
e
1
+ k

v
1
e
1
+ k

p
1
e
1
= 0
e
2
+ k

v
2
e
2
+ k

p
2
e
2
= 0
where e
i
is the error at joint i, e
i
= (
i

id
). Now, we need to choose k

v
i
and k

p
i
to achieve critical damping, and to achieve our desired closed-loop frequencies. For a
unit-mass system, we choose
k

p
i
=
2
i
k

v
i
= 2
i

i
So, we get
k

p
1
= 100, k

v
1
= 20
k

p
2
= 400, k

v
2
= 40
(d) If
2
= 180

, what is the steady-state error vector for a given disturbance

torque,
dist
= [2 4]
T
?
The controlled system, with a disturbance torque
dist
is
M()

+ V (

, ) = +
dist
Substituting in our form for =

+ yields
M()

M()

=
dist
This has the form
M()
_
e + K

v
e + K

p
e
_
=
dist
where e is the error vector e =
d
, and K

v
and K

p
are the matrices given by
K

v
=
_
k

v
1
0
0 k

v
2
_
, K

p
=
_
k

p
1
0
0 k

p
2
_
In the steady state (e = e = 0), the equation is
M()K

p
e =
dist
which means that the steady state error is
e = (M()K

p
)
1

dist
For our values, this is:
e =
__
3 0
0 1
_ _
100 0
0 400
__
1
_
2
4
_
=
_
300 0
0 400
_
1
_
2
4
_
=
_
1
150
1
100
_