MANIPULATOR DYNAMICS

Readings
• Chapter 6 of Introduction to Robotics – Mechanics and Control, J.J.
Craig (any edition).

Acknowledgement: In preparing this presentation, I have taken a few

slides from R. Wood’s lectures.
Past coverage
 We learned systematic methods to determine the forward and
inverse kinematics (for any serial manipulator):
 Forward kinematics: mapping from joint variables to the Cartesian
position and orientation of the end effector
 Inverse kinematics: finding joint variables that satisfy a given Cartesian
position and orientation of the end effector
 Jacobian: mapping from the joint velocities to the end effector linear and
angular velocities
 The next step is determining the dynamics (for any serial
manipulator).
 Dynamics: definition
 Dynamics is the set of equations of motion for any n-DOF
system
 In general, this will consist of n coupled, second-order ODEs
 We can write the dynamics for systems that are
 Linear or nonlinear
 Conservative or non-conservative (i.e., dissipative)
 Why do we need the robot dynamics?
 The dynamical model of a robot can be used in choosing an
appropriate controller for reaching a desired outcome in terms of
robot position and/or force.

Linear system :

·
Superposition property
Dynamics of linear systems vs. nonlinear systems
 Write the dynamics of
f mjj
 A point mass under no external force =

 A free-swinging pendulum

m((8) + my sin = 0

+ sin0 = 0
in
.......
-- ---- to sino
my
Dj
↓
..
,
-

my
 Two method for deriving a robot’s dynamics
 Newton-Euler formulation
 This is a “force-balance” approach which is iterative and is based on
Newton’s equation and Euler’s equation
 Newton’s 2nd Law of Motion: F  mi C

 Euler’s Angular Force/Moment equation: N moment  I CM i i  i  I CM i i

 Iterates from base to end-effector to find velocities and accelerations

 Iterates from end-effector to base to compute forces and moments
 Lagrangian formulation
 This is an “energy-based” approach in which what we need to know are the
conservative (kinetic and potential) and non-conservative (dissipative such
as friction) terms
 We will only cover the Lagrangian method.
Example: 1-DOF point mass system dynamics
 Let us derive the equations of motion for a 1-DOF mass
 Consider a particle of mass m
 Newton’s second law: my  f  mg

 We know the kinetic and potential energies:

1 2
K my U  mgy
2
U
 Re-write the first differential equation
 Left side:

my 
d
my   d   1 my 2   d K
dt dt y  2  dt y
 Right side:

mg  mgy  U
y y
Example: 1-DOF point mass system dynamics
 So the first differential equation can be re-written as
d K U
 f
dt y y

La
 Therefore, if we make the following definition:
L  K U
where L is called the Lagrangian, the equation of motion
for the 1-DOF point mass becomes (why?)
d L L
  f O for this
dt y y .

·
by f &
by
+
Ou
by
=
F

U isn't function of
a if
 The Lagrangian method for writing the dynamics of a
robot does the same thing backward.
 Uci : linear velocity COM of link i

Lagrangian formulation · of

· "Wi :
angular velocity of linki as expressed in Ei
 The equations of motion for an n-DOF system is
& Cis attached to link;
·
Imagine trace a

d L L at its COM and having the same

  Orientation as frame Ei3

dt   · Pi : Vector locating the COM of link ; in 203
·
og :
gravity vector
 This leads to a set of n coupled, 2nd-order differential equations
 Kinetic energy of a manipulator: Sum of kinetic energy of all of its links
n n
1 1i T
K   ki  mi v C v C  i
T Ci
I i ii
i 1 2 2
i i
i 1
↑

 Potential energy of a manipulator: Sum of potential energy of all of its links

n n
U   ui   mi 0g T 0PCi urefi
i 1 i 1
Example: RP manipulator dynamics
 Derive the dynamics of the RP manipulator for which the inertia V od
(620) +

tensors of the two links are given as ..... da

/
di O, --
---

Known
Va = L,
0 ,

-
>
↑ ↳ Eas

not
-

Eis to-
T-
·

known
-
> ~ da sinO ,

 Solution:
o
(08) 'w "I w = F+z
'w = = ,
, ,

Izz0
(8) n.
· =
wn = =
 Example: RP manipulator dynamics

of o Oto
-
= +

↓ Equivalent
&

= T
Example: RP manipulator dynamics

E
-
⑩ D
T
M(0)8 v(0, 8) + G(0) =

( =)
+ -

(8)
 Example: PR manipulator dynamics
Inertin tensor not given C 0
 Exercise 6.16 of Craig: Derive the dynamics of this PR manipulator.

Figure 6.10
↓9 PR manipulator of
di
Exercise 6.16.

CEL
--
>

GE (O)
& ·

REMOKEE
0 u = 0
U 0
, Uz =
, =

)=
-
=
(smitmust
Example: PR manipulator dynamics
 Example: PP manipulator dynamics
Consider the following planar manipulator, which consists of two links and two prismatic joints (the first
link moves up and down, and the second link moves left and right). Assume that the center of mass of
each link is somewhere on that link (exactly where on the link does not matter in this problem). Denote
the masses of the two links by m1 and m2, respectively, and denote the displacement of the two prismatic
joints (i.e., the linear travel of the two links) by q1 and q2, respectively.

O
Von =
q
-2
n
Constan
>
⑳-

=
v
-----,

Using the Lagrangian formulation, find the dynamics of the above robot. Note that an answer
provided through inspection without using the Lagrange’s formula is not enough. When using the
Lagrange’s method, you can assume that the ground level shown above is where the potential energy
is zero.
 Example: PP manipulator dynamics
↳, = , Eme(q +)
m
2
=

U = .
g, 2 = M29( ,
+ Content) irrelevant

k = [( ,
+ mz)q2 + zm2q

U = (m , mz) 99
+ ,

- 00 00 + = 5

↓ ↓
((m the f i
,
,
jo (( ,

+mn)9)

Emmette
in

(i) (ts) () =
Example: Va
E
=ju +

9t >
(2)
Oz

di

knowledge of formary
Rinematics
T

x2
- yu
-- -
 Coriolis & centrifugal Vector gravity vector

T ↑
T
m(0) v(0 , 8) G(0)
W
Innertial
matrix