Lecture Notes W2

ROBOTICS : BASIC AND A DVANCED C ONCEPTS
M ATHEMATICAL P RELIMINARIES , D-H C ONVENTION & E XAMPLES
Ashitava Ghosal
Department of Mechanical Engineering

&
Centre for Product Design and Manufacturing
Indian Institute of Science
Bangalore 560 012, India
Email: asitava@iisc.ac.in
NPTEL, 2020
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
A SHITAVA G HOSAL (IIS C ) ROBOTICS : BASIC AND A DVANCED C ONCEPTS NPTEL, 2020 1 / 70
1 C ONTENTS
2 L ECTURE 1
Mathematical Preliminaries
Homogeneous Transformation
3 L ECTURE 2
Elements of a robot – Joints
Elements of a robot – Links
4 L ECTURE 3
Examples of D-H Parameters & Link Transformation Matrices
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O UTLINE
1 C ONTENTS
2 L ECTURE 1
3 L ECTURE 2
4 L ECTURE 3
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P OSITION OF A R IGID B ODY
ẐA
{A} Right-handed coordinate system
specified by
Origin OA .
A
p Set of 3 mutually orthogonal axis
— Unitvectors X̂A , ŶA and ẐA are
along the index finger, the middle
finger and the thumb of the
OA ŶA
right-hand, respectively.
Label to keep track — {A}.
Rigid Body A
X̂A
Position of a point of interest such as
centre of mass/gravity
F IGURE : Position of point P denoted by A p
Point A p with Cartesian coordinates (px , py , pz )T

A
p = px X̂A + py ŶA + pz ẐA = (px , py , pz ).T . .
(1).
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O RIENTATION OF A R IGID B ODY
Position of one point on the rigid body not enough to describe it in 3D space.
Orientation of a rigid body B with respect to {A}
ẐA
k̂
{A}
ẐB
{B} φ
B
p Attach coordinate system, {B},
ŶB
to rigid body B.
Origin of {A} and {B}
OA ,OB ŶA coincident.
Obtain description of {B} with
respect to {A}.
Rigid Body B
X̂A X̂B
F IGURE : Orientation of a rigid body . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .
O RIENTATION – D IRECTION C OSINES
Unit vectors X̂B , ŶB , and ẐB , attached to B, can be described in {A}
A
X̂B = r11 X̂A + r21 ŶA + r31 ẐA
A
ŶB = r12 X̂A + r22 ŶA + r32 ẐA (2)
A
ẐB = r13 X̂A + r23 ŶA + r33 ẐA
rij , i, j = 1, 2, 3 are called direction cosines

r11 =A X̂B · X̂A = |A X̂B | · |X̂A | cos(A X̂B , X̂A ) = cos(A X̂B , X̂A )
Define 3 × 3 rotation matrix A
B [R] with rij , i, j = 1, 2, 3 as its elements
The first column of A B [R], [r11 , r21 , r31 ] , is same X̂B →
T A
A [R] = [A X̂ | A Ŷ | A Ẑ ].
B B B B
A [R] completely describes all
B three coordinate axis of {B} with respect to {A}
=> A B [R] gives orientation of rigid body B in {A}.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O RIENTATION – P ROPERTIES OF A
B [R]
The vector B p in rigid body B can be described in {A} by

A
p=A B
B [R] p (3)
A [R] is orthonormal ⇒ |A X̂ | = |A Ŷ | = |A Ẑ | = 1, and
B B B B
A X̂ ·A Ŷ =A Ŷ ·A Ẑ =A Ẑ ·A X̂ = 0 => only three independent
B B B B B B
parameters out of 9 rij ’s.
det(A
B [R]) = +1
A [R]T A [R] = [U], where [U] is a 3 × 3 identity matrix.
B B
∆ −1 T
Inverse is same as transpose — B A
A [R] = B [R] =A
B [R] .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O RIENTATION – P ROPERTIES OF A
B [R]
A [R] is a 3 × 3 matrix
B
±ıϕ , where ı =
√
Three eigenvalues of A B [R] are +1, e −1, and
−1 r11 +r22 +r33 −1
ϕ = cos ( 2 ).
The eigenvector corresponding to +1 is
k̂ = (1/2 sin ϕ )[r32 − r23 , r13 − r31 , r21 − r12 ]T , ϕ ̸= {0, nπ }, where n is a
natural number. For ϕ = {0, 2nπ }, there is no rotation and ϕ = 2(n − 1)π
needs to handled specially!
Rotation axis k̂ fixed in {A} and {B}:
A k̂
=A B B
B [R] k̂ = 1 k̂.
Follows from transformation of a vector from {B} to {A} and the definition of
an eigenvector.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O RIENTATION U SING (k̂, ϕ )
B [R] in terms of (kx , ky , kz ) and angle ϕ

Elements of A T
r11 = kx 2 (1 − cos ϕ ) + cos ϕ

r12 = kx ky (1 − cos ϕ ) − kz sin ϕ
r13 = kz kx (1 − cos ϕ ) + ky sin ϕ
r21 = kx ky (1 − cos ϕ ) + kz sin ϕ
r22 = ky 2 (1 − cos ϕ ) + cos ϕ (4)
r23 = ky kz (1 − cos ϕ ) − kx sin ϕ
r31 = kz kx (1 − cos ϕ ) − ky sin ϕ
r32 = ky kz (1 − cos ϕ ) + kx sin ϕ
r33 = kz 2 (1 − cos ϕ ) + cos ϕ
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O RIENTATION – S IMPLE ROTATION
Rotation axis k̂ is parallel to X̂A and hence to X̂B → Rotation about X̂ axis.
 
1 0 0
ẐA
B [R] = [R(X̂, ϕ )] =
A  0 cos ϕ − sin ϕ  (5)
{A} 0 sin ϕ cos ϕ
ẐB
{B}
B
p
ŶB
Rotation about X̂ shown in
φ
OA , OB ŶA
Figure.
k̂
φ
Rigid Body B
X̂A , X̂B
F IGURE : Rotation about X̂ by ϕ . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .
S IMPLE ROTATIONS (C ONTD .)
Rotation about Ŷ and Ẑ

 
cos ϕ 0 sin ϕ
[R(Ŷ, ϕ )] =  0 1 0  (6)
− sin ϕ 0 cos ϕ
 
cos ϕ − sin ϕ 0
[R(Ẑ, ϕ )] =  sin ϕ cos ϕ 0  (7)
0 0 1
Rotation matrices in equations (5) through (7) called simple rotations.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
S UCCESSIVE ROTATIONS
Two successive rotations:
1 Initially B is coincident with {A}.
2 First rotation relative to {A}. After first rotation {A} → {B1 }.
3 Second rotation relative to {B1 }. After second rotation {B1 } → {B}.
ẐA
{A}
Resultant rotation:
ẐB
{B} A [R] = A [R] B1 [R] — Note order of
ŶB1 B B1 B
ẐB1 matrix multiplication.
ŶB
Resultant of n rotations —
A [R] = A [R] B1 [R] ... Bn−1 [R]
OA ,OB1 ,OB ŶA B B1 B2 B
Matrix multiplication is non
X̂A Rigid Body B
commutative in general —
X̂B A [R] B1 [R] ̸= B1 [R] A [R]
X̂B1 Rigid Body B1 B1 B B B1
⇒ Order of rotation is important!
F IGURE : Successive rotation
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O RIENTATION – T HREE A NGLES
Orientation described by 3 independent parameters → Three rotations

completely describe orientation of a rigid body.
Three successive rotations about axes fixed to moving body
Rotations about three distinct axes: 6 combinations – X-Y-Z, X-Z-Y, Y-Z-X,
Y-X-Z, Z-X-Y & Z-Y-X
Rotations about two distinct axes: 6 combinations – X-Y-X, X-Z-X, Y-X-Y,
Y-Z-Y, Z-X-Z, & Z-Y-Z
Rotations can also be about axes fixed in space – 12 possible combinations for
3 and 2 distinct axes.
Minimal representation of orientation of rigid body – Only three parameters
(angles) and no constraints.
3 angles about two distinct axes are called classical Euler angles and about 3
distinct axes are also called Tait-Bryan angles.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
X–Y–Z E ULER ANGLES  
1 0 0
Rotation about X̂ – A [R] = [R(X̂, θ )] = 
0 cos θ1 − sin θ1 
B1 1
0 sin θ1 cos θ1
 
cos θ2 0 sin θ2
Rotation about Ŷ – B  
B2 [R] = [R(Ŷ, θ2 )] =
1
0 1 0
− sin θ2 0 cos θ2
X̂A , X̂B1 X̂B2 X̂A , X̂B1 X̂B2 X̂A , X̂B1
{A} {A}, {B1 }
{B1 } {A}, {B1 } {B2 }
{B2 } X̂B
θ1 ẐB2 ẐB , ẐB2
ẐB1 {B} θ3
ẐB1 ẐB1
OA , OB ẐA OA , OB ẐA OA , OB ẐA
ŶA ŶA θ2 ŶA

ŶB1 ŶB
ŶB1 , ŶB2 ŶB1
F IGURE : X–Y–Z Euler angles . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .
X–Y–Z E ULER ANGLES (C ONTD .)
 
cos θ3 − sin θ3 0
 sin θ3 cos θ3 0 
Rotation about Ẑ – B B [R] = [R(Ẑ, θ3 )] =
2
0 0 1
Resultant rotation –  
c2 c3 −c2 s3 s2
A [R] = A [R] B1 [R] B2 [R] =  s s c + s c −s1 s2 s3 + c3 c1 −s1 c2 
B B1 B2 B 1 2 3 3 1
−c1 s2 c3 + s3 s1 c1 s2 s3 + c3 s1 c1 c2
Note: ci , si denote cos θi and sin θi , respectively.
A [R]
B gives orientation of {B} given X–Y–Z angles.
A [R] will be different if order of rotations is different.
B
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
X–Y–Z E ULER ANGLES (C ONTD .)
Given A
B [R], find X–Y–Z Euler angles.
Algorithm rij ⇒ θi for X–Y–Z rotations
If r13 ̸= ±1, then √
θ2 = atan2(r13 , ± (r11 2 + r 2 ))
12
θ1 = atan2(−r23 / cos θ2 , r33 / cos θ2 )
θ3 = atan2(−r12 / cos θ2 , r11 / cos θ2 )
Else
If r13 = 1, θ1 = atan2(r21 , r22 ), θ2 = π /2, θ3 = 0
If r13 = −1, θ1 = −atan2(r21 , r22 ), θ2 = −π /2, θ3 = 0
atan2(y , x): four-quadrant arc-tangent function (see function atan2 in
MATLAB⃝ R
) – θ1 , θ2 , θ3 ∈ [−π , π ].
Two sets of values of θ1 , θ2 and θ3 .
θ2 = ±π /2 → θ1 , θ3 not unique – θ1 ± θ3 can be found.
Singularities in Euler angle representation. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
Z–Y–Z E ULER ANGLES
ẐA , ẐB1 ẐB2 , ẐB
ẐB2 ẐA , ẐB1 ẐA , ẐB1
{A} {A}
{B1 } {A}, {B1 } {B}
{B2 }
θ1 θ3 ŶB
ŶB1 ŶB1 , ŶB2
θ2
ŶB1 , ŶB2
OA , OB ŶA OA , OB ŶA OA , OB ŶA
X̂A
X̂A X̂A
X̂B1 X̂B1 X̂B2 X̂B1 X̂B

X̂B2
F IGURE : Z–Y–Z Euler angles
   
c1 −s1 0 c2 0 s2 c3 −s3 0
A [R] =  s1 c1 0  0 1 0   s3 c3 0 
B
 0 0 1 −s2 0 c2 0
 0 1
(8)
c1 c2 c3 − s1 s3 −c1 c2 s3 − s1 c3 c1 s2
=  s1 c2 c3 + c1 s3 −s1 c2 s3 + c1 c3 s1 s2 
−s2 c3 s2 s3 c2 . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
Z–Y–Z E ULER ANGLES (C ONTD .)
Algorithm rij ⇒ Z–Y–Z Euler angles
If r33 ̸= ±1, then √
θ2 = atan2(± (r31 2 + r 2 ), r )
32 33
θ1 = atan2(r23 / sin θ2 , r13 / sin θ2 )
θ3 = atan2(r32 / sin θ2 , −r31 / sin θ2 )
Else
If r33 = 1, then
θ1 = θ2 = 0, θ3 = atan2(−r12 , r11 )
If r33 = −1, then
θ1 = 0, θ2 = π , θ3 = atan2(r12 , −r11 )
Two possible sets of Z–Y–Z Euler angles (θ1 , θ2 , θ3 ) for a given A
B [R].
r33 = ±1 → Singularity → θ1 ± θ3 can be found.
For unique θ1 and θ3 when r33 = ±1 → Choose θ1 = 0.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
OTHER R EPRESENTATIONS
Euler parameters (see Kane et al., 1983, McCarthy, 1990): 4 parameters

derived from k̂ = (kx , ky , kz )T and angle ϕ
3 parameters — ε = k̂ sin ϕ /2
fourth parameter — ε4 = cos ϕ /2
One constraint — ε1 2 + ε2 2 + ε3 2 + ε4 2 = 1
Quaternions (pp. 185 Arfken(1985), Wolfram website): 4 parameters
‘Sum’ of a scalar q0 and a vector (q1 , q2 , q3 )T — Q = q0 + q1 î + q2 ĵ + q3 k̂.
î, ĵ, and k̂ are the unit vectors in ℜ3 .
Product of two quaternion also a quaternion.
Inverse of quaternion exists.
q02 + q12 + q22 + q32 = 1 → Unit quaternion represent orientation of a rigid body in
ℜ3 .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O RIENTATION OF A R IGID B ODY –
S UMMARY
Orientation of a rigid body in ℜ3 is specified by 3 independent parameters.

Various representation of orientation with their own advantages and
disadvantages
Rotation matrix A B [R] – 9 rij ’s + 6 constraints → Too many variables and
constraints but ideal for analysis.
Axis (kx , ky , kz )T and angle ϕ – 4 parameters + one constraint kx2 + ky2 + kz2 = 1
→ Useful for insight and extension to screws, twists and wrenches.
Euler angles: 3 parameters and zero constraints→ Minimal representation but
suffer from problem of singularities and sequence must be known.
Euler parameters and quaternions: 4 parameters +1 constraint → Similar but
not exactly same as axis-angle form, no singularities, used in motion planning.
Can convert from one representation to any other for regular cases!
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
C OMBINED T RANSLATION AND
O RIENTATION OF A R IGID B ODY
ẐA Rigid Body B
{A} ẐB
{B} B {A} and {B}, OA and OB not
p
coincident
OB
A
ŶB Orientation of {B} with respect
OB
A
p to {A} – AB [R]
Ap =A OB + A B
B [R] p
AO locates OB with respect to
OA ŶA B
OA .
X̂B
X̂A
F IGURE : General transformation

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O UTLINE
1 C ONTENTS
2 L ECTURE 1
3 L ECTURE 2
4 L ECTURE 3
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
4 × 4 T RANSFORMATION M ATRIX
Combined translation and orientation

Define A P = [A p | 1]T and B P = [B p | 1]T → A P and B P are called 4 × 1
homogeneous coordinates
Rewrite A p =A OB + A B
B [R] p and 1 = 1 as
( A [R] AO
)
A B B
P= B P=A B
B [T ] P (9)
0 0 0 1
A [T ]
B is the 4 × 4 homogeneous transformation matrix
In computer graphics and computer vision → Last row is used for perspective
and scaling and not [0 0 0 1].
Upper left 3 × 3 matrix is identity matrix → Pure translation.
Top right 3 × 1 vector is zero → Pure rotation.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
4 × 4 T RANSFORMATION M ATRIX –
P ROPERTIES
( )
A [R]T −A
TA
Inverse of A [T ] is given by A [T ]−1 = B B [R] OB
B B
0 0 0 1
B1
Two successive transformations {A} → {B1 } → {B} gives A A
B [T ] = B1 [T ] B [T ].
( A [R]B1 [R] A [R]B1 O + A O
)
A [T ] = B1 B B1 B B1
B
0 0 0 1
B1 Bn−1
n successive transformations A A
B [T ] = B1 [T ] B2 [T ] ... B [T ]
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P ROPERTIES OF A
B [T ] (C ONTD .)
Four eigenvalues of A ±ıϕ , ϕ same as for A [R].

B [T ] are +1, +1, e B
Eigenvectors for +1 is k̂ – No other eigenvector!
represents the general motion of a rigid body in 3D space → 6
A [T ]
B
parameters must be present.
General motion of rigid body as a twist – Rotation about a line and
translation along the line.
Direction of the line: (kx , ky , kz )T
T
([U ]−A A
B [R ] ) OB
Location of the line in ℜ3 : (k̂, Y × k̂) where Y = 2(1−cos ϕ )
→ 4 independent parameters in k̂ and Y × k̂.
Rotation about the line: angle ϕ → 1 independent parameter.
Translation along the line: (A OB · k̂) → 1 independent parameter.
More details about twists, eigenvalues and eigenvectors (see Sangamesh
Deepak and Ghosal, 2006).
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
S UMMARY
Definition of a right-handed coordinate system {A} → X̂, Ŷ, Ẑ and origin OA .
A rigid body B in 3D space has 6 DOF with respect to another rigid body A: 3
for position & 3 for orientation → Rigid body B conceptually identical to a
coordinate system {B}.
Position of rigid body → Position of a point of interest on rigid body with
respect to coordinate system {A} → 3 Cartesian coordinates:
A p = (p , p , p )T .
x y z
B [R], 2) (ϕ , k̂)
Orientation described in many ways: 1) by 3 × 3 rotation matrix A
or angle-axis form, 3) 3 Euler angles, 4) Euler parameters & quaternions.
Algorithms available to convert one representation to another.
4 × 4 homogeneous transformation matrix, A
B [T ], represent position and
orientation in a compact manner.
Properties of A
B [T ] can be related to a screw.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O UTLINE
1 C ONTENTS
2 L ECTURE 1
3 L ECTURE 2
4 L ECTURE 3
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
J OINTS – I NTRODUCTION
A joint connects two or more links.

A joint imposes constraints on the links it connects.
2 free rigid bodies have 6 + 6 degrees of freedom.
Hinge joint connecting two free rigid bodies → 6 + 1 degrees of freedom.
Hinge joint imposes 5 constraints, i.e., hinge joint allows 1 relative (rotary)
degree of freedom.
Degree of freedom of a joint in 3D space: 6 − m where m is the number of
constraint imposed.
Serial manipulators → All joints actuated → One-degree-of-freedom joints used.
Parallel and hybrid manipulators → Some joints passive →
Multi-degree-of-freedom joints can be used.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
T YPES OF J OINTS DOF
F IGURE : – Types of joints . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .
C ONSTRAINTS IMPOSED BY A k̂
{i}
ROTARY (R) JOINT Rigid Body i − 1 θi
Rigid bodies, {i − 1} and {i}, Oi
connected by a rotary (R) joint.

i
p
{i − 1}
{i} can rotate about k̂, with respect Oi−1

i−1
p
to {i − 1}, by angle θi
Rigid Body i
0 [R] = 0 [R] i−1 [R(k̂, θ )]
i i−1 i i
Three independent equations in the
0
Oi Li Rotary Joint
0
matrix equation above; θi is

Oi−1 0
p
unknown⇒ 2 constraints in the 3 {0}
equations.
F IGURE : A rotary joint

For a common point P on the rotation axis along line Li
0 p =0 O 0 i−1 p =0 O + 0 [R]i p ⇒ 3 constraints present.
i−1 + i−1 [R] i i
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
C ONSTRAINTS IMPOSED BY A k̂
{i}
ROTARY (R) JOINT IN A LOOP Rigid Body i − 1 θi
Oi
i
p
{i − 1}
i−1
p
Oi−1 R
Oi
Rotary joint in a loop — 2 Rigid Body i
ends {L} and {R}.

Li Rotary Joint
L
Oi−1 {R}
{L}
L
D
F IGURE : A rotary joint in a loop

2 constraints: L [R]
i = i−1 [R] i [R(k̂, θi )] = LR [R]R
L i−1
i [R], θi unknown.
Three constraints in L p =L Oi−1 + Li−1 [R]i−1 p =L D +R Oi + R
. i. [R]
i
. . . .p. . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
C ONSTRAINTS IMPOSED BY A
SPHERICAL (S) JOINT {i}
Rigid Body i − 1
i
p
{i − 1} i−1
p
Spherical(S) or ball and socket

joint allows three rotations. Rigid Body i
S joint can be represented as 3 0

Oi−1 0
p
0
Oi Spherical Joint
intersecting rotary(R) joints.

{0}
F IGURE : A spherical joint

3 constraints: 0p =0 Oi−1 + 0i−1 [R]i−1 p =0 Oi + 0i [R]i p . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
C ONSTRAINTS IMPOSED BY A
Rigid Body j
SPHERICAL - SPHERICAL (S–S) JOINT {j}
Sj
PAIR Rigid Body i j
Sj
Si lij
{i}
i
Si
The S–S pair appear in

S-S Pair
many parallel manipulators. R
Sj R
O
Distance between two

spherical joint is constant. L
Oi
L
Si
{L}
L
D
F IGURE : A S–S pair in a loop

Constraint: (L Si − (L D +R Sj )) · (L Si − (L D +R Sj )) = lij2 , lij. is
. . a. .constant.
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O UTLINE
1 C ONTENTS
2 L ECTURE 1
3 L ECTURE 2
4 L ECTURE 3
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
L INKS – I NTRODUCTION
A link A is a rigid body in 3D space → Can be described by a coordinate

system {A}.
A rigid body in ℜ3 has 6 degrees of freedom → 3 rotation + 3 translation → 6
parameters
For links connected by rotary (R) and prismatic (P), possible to use 4
parameters – Denavit-Hartenberg (D-H)parameters (see Denavit & Hartenberg,
1955).
4 parameters since lines related to rotary(R) and prismatic (P) joint axis are
used.
For multi-degree-of-freedom joints → Use equivalent number of
one-degree-of-freedom joints.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
L INKS – D ENAVIT-H ARTENBERG
(D-H) PARAMETERS
A word about conventions – Several ways to derive D-H parameters!

Convention used:
1 Coordinate system {i} is attached to the link i.
2 Origin of {i} lies on the joint axis i – Link i is “after” joint i.
3 “after” for serial manipulators – Numbers increasing from fixed {0} → Link 1 {1}
→ . . . → Free end {n}.
4 “after” for parallel manipulators – Not so straight-forward due to one or more
loops.
Convention same as in Craig (1986) or Ghosal(2006).
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
A SSIGNMENT OF COORDINATE AXES
Three intermediate links — {i − 1}, {i} and {i + 1}.
Ẑi+1
Rotary joint axis — Labeled Ẑi−1 , Ẑi and Ẑi+1 . X
Link i
Ẑi−1 Ẑi {i}
θi
{i − 1}
Oi
Link i + 1
Link i − 1
Link i − 2 di
αi−1 Li+1
Ẑ0 X̂i−1
{0} Li
Oi−1 ai−1
Li−1
Ŷ0
X̂0
. . . . . . . . . . . . . . . . . . . .
F IGURE : Intermediate links and D-H parameters
. . . . . . . . . . . . . . . . . . . .
D-H PARAMETERS – A SSIGNMENT
OF COORDINATE AXIS (C ONTD .)
For coordinate system {i − 1}

Ẑi−1 is along joint axis i − 1.
X̂i−1 is chosen along the common perpendicular between lines Li−1 and Li .
Ŷi−1 = Ẑi−1 × X̂i−1 – Right-handed coordinate system.
The origin Oi−1 is the point of intersection of the mutual perpendicular line and
the line Li−1 .
For coordinate system {i}: Ẑi is along the joint axis i, X̂i is along the common
perpendicular between Ẑi and Ẑi+1 , and the origin of {i}, Oi , is the point of
intersection of the line along X̂i and line along Ẑi .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
D-H PARAMETERS FOR LINK i
Twist angle αi−1 — Twist angle for link i has subscript i − 1!
Angle between joints axis i − 1 and i & measured about X̂i−1 using right-hand
rule.
Signed quantity between 0 and ±π radians.
Link length ai−1 — Link length for link i is ai−1 !
Distance between joints axis i − 1 and i & measured along X̂i−1 .
Always a positive quantity.
Link offset di
Measured along Ẑi from X̂i−1 to X̂i – Can be < 0.
Joint i rotary → di constant.
Joint i prismatic → di is joint variable.
Rotation angle θi
Angle between X̂i−1 and X̂i measured about Ẑi using right-hand rule – Between
0 and ±π radians.
Joint i is prismatic → θi is constant.
Joint i rotary → θi is joint variable. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
D-H PARAMETERS – S PECIAL CASES
Consecutive joints axis parallel → ∞ common perpendiculars
αi−1 = 0, π , ai−1 along any common perpendicular.
Joint i is rotary (R) → di is taken as constant (often zero).
Joint i is prismatic (P) → θi is taken as zero.
Consecutive P joints parallel → d’s not independent!
Consecutive joints axis intersecting → ai−1 = 0 and X̂ normal to plane.
First link {1}: choice of Ẑ0 and thereby X̂0 is arbitrary.
R joint → Choose {0} and {1} coincident & αi−1 = ai−1 = 0, di = 0.
P joint → Choose {0} and {1} parallel & αi−1 = ai−1 = θi = 0.
Last link {n}: Ẑn+1 not defined
R joint → Origins of {n} and {n + 1} are chosen coincident & dn = 0 and θn = 0
when X̂n−1 aligns with X̂n .
P joint → X̂n is chosen such that θn = 0 & origin On is chosen at the
intersection of X̂n−1 and Ẑn when dn = 0.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
L INK TRANSFORMATION MATRICES
Four D-H parameters describe link i with respect to link i − 1
Orientation of of {i} with respect to {i − 1} is given by
i [R] = [R(X̂, αi−1 )][R(Ẑ, θi )]
i−1
Location of origin of {i} with respect to {i − 1} is given by

i−1 O = a i−1 X̂ i−1 Ẑ
i i−1 i−1 + di i
i−1
Recall i−1 X̂i−1 = (1, 0, 0)T . The vector i−1 Ẑ
i is the last column of i [R] and
is (0, −sαi−1 , cαi−1 )T 1 .
The 4 × 4 transformation matrix relating {i} with respect to {i − 1} is
 
cθi −sθi 0 ai−1
 sθ cα 
i−1  i i−1 cθi cαi−1 −sαi−1 −sαi−1 di  (10)
i [T ] =  s s cαi−1 di 
θi αi−1 cθi sαi−1 cαi−1
0 0 0 1
. . . . . . . . . . . . . . . . . . . .
1 The symbols −sαi−1 , cαi−1 denote sin(αi−1 ) and cos(αi−1 ), respectively
. . . . . . . . . . . . . . . . . . . .
L INK TRANSFORMATION MATRICES
i−1
i [T ] is a function only one joint variable — di or θi .
i−1 O locates a point on the joint axis i at the beginning of link i.
i
Position and orientation of link i determined by αi−1 and ai−1 . Note: subscript
i − 1 in the twist angle and length!
The mix of subscripts are a consequences of the D-H convention used!
Link i with respect to {0} — 0i [T ] = 01 [T ] 12 [T ] ... i−1
i [T ]
The position and orientation of any link can be obtained with respect any other
link by suitable use of 4 × 4 link transformation matrices.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
S UMMARY
Two mechanical elements of robots – Links and joints.
Joints allow relative motion between connected links → Joints impose
constraints.
Serial robots mainly use one-degree-of-freedom rotary (R) and prismatic (P)
actuated joints.
Parallel and hybrid robots use passive multi-degree-of-freedom joints and
actuated one-degree-of-freedom joints.
One degree-of-freedom R and P joints represented by lines along joint axis —
Ẑ is along joint axis.
Formulation of constraints imposed by various kinds of joints.
Link is a rigid body in 3D space → Represented by 4 Denavit-Hartenberg
(D-H) parameters.
Convention to derive D-H parameters and for special cases.
4 × 4 link transformation matrix in terms of D-H parameters.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
O UTLINE
1 C ONTENTS
2 L ECTURE 1
3 L ECTURE 2
4 L ECTURE 3
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
S ERIAL M ANIPULATORS
Steps to obtain D-H parameters for a n link serial manipulator

1 Label joint axis from 1 (fixed) to n free end.
2 Assign Ẑi to joint axis i, i = 1, 2, ..., n.
3 Obtain mutual perpendiculars between lines along Ẑi−1 and Ẑi – X̂i−1 .
4 Origin Oi−1 on joint axis i − 1.
5 Handle special cases – a) consecutive joint axes parallel and perpendicular, b)
first and last link
6 Obtain 4 D-H parameters for link i, αi−1 , ai−1 , di and θi .
i−1
For each link i obtain i [T ] using equation (10).
Obtain link transform for link i with respect to fixed coordinate system {0},
0 [T ],
by matrix multiplication.
i
Obtain, if required, ij [T ] by appropriate matrix operations.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P LANAR 3R M ANIPULATOR –
A SSIGNMENT OF COORDINATE
X̂3 , X̂T ool
SYSTEMS X̂2
{T ool}
Link 3 θ3
ŶT ool l3
All 3 rotary joint axis Ẑi parallel O3
and pointing out. {3}

Link 2
Ŷ3
{0} — Ẑ0 is pointing out, X̂0 Ŷ0
Ŷ1 l2
and Ŷ0 pointing to the right and Ŷ2 {2} θ2 X̂1
top, respectively. {0}
Origin O0 is coincident with O1 {1}

Link 1
O2
l1
shown in figure. θ1
X̂0
O1
F IGURE : The planar. 3R manipulator

. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P LANAR 3R M ANIPULATOR –
A SSIGNMENT OF COORDINATE
SYSTEMS
For {1} origin O1 and Ẑ1 are coincident with O0 and Ẑ0 .
X̂1 and Ŷ1 are coincident with X̂0 and Ŷ0 when θ1 is zero.
X̂1 is along the mutual perpendicular between Ẑ1 and Ẑ2 .
X̂2 is along the mutual perpendicular between Ẑ2 and Ẑ3 .
For {3}, X̂3 is aligned to X̂2 when θ3 = 0.
O2 is located at the intersection of the mutual perpendicular along X̂2 and Ẑ2 .
O3 is chosen such that d3 is zero.
The origins and the axes of {1}, {2}, and {3} are shown in Figure.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P LANAR 3R M ANIPULATOR – D-H
TABLE
From the assigned axes and origins, the D-H parameters are as follows:
i αi−1 ai−1 di θi
1 0 0 0 θ1
2 0 l1 0 θ2
3 0 l2 0 θ3
l1 and l2 are the link lengths and θi , i = 1, 2, 3 are rotary joint variables.
Length of the end-effector l3 does not appear in the table — Recall: D-H
parameters are till the origin which is at the beginning of a link!
For end-effector frame, {Tool}:
Axis of {Tool} parallel to {3} → 3Tool [R] is identity matrix.
Origin of {Tool} at the mid-point of the parallel jaw gripper, at a distance of l3
from O3 along X̂3 .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P LANAR 3R M ANIPULATOR – L INK
TRANSFORMS
Link transformation
 matrix (repeated): 
cθi −sθi 0 ai−1
 
i−1  sθi cαi−1 cθi cαi−1 −sαi−1 −sαi−1 di 
i [T ] =  s s 
θi αi−1 cθi sαi−1 cαi−1 cαi−1 di
0 0 0 1
 
c1 −s1 0 0
 s1 c1 0 0 
Substitute first row of D-H table (1 0 0 θ1 ) – 01 [T ] = 
 0

0 1 0 
0 0 0 1
 
c2 −s2 0 l1
 s2 c2 0 0 
The second row of the D-H table (0 l1 0 θ2 ) – 12 [T ] =  0

0 1 0 
0. . . 0. . . . 0. . . 1. . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P LANAR 3R M ANIPULATOR : L INK
TRANSFORMS (C ONTD .)
 
c3 −s3 0 l2
 s3 c3 0 0 
The third row of D-H table (0 l2 0 θ3 ) – 23 [T ] = 
 0

0 1 0 
0 0 0 1
 
1 0 0 l3
 0 1 0 0 
The 3Tool [T ] is obtained as 3Tool [T ] = 
 0 0 1 0 

0 0 0 1
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
P LANAR 3R M ANIPULATOR : L INK
To obtain 0 0 1 2
 3 [T ] multiply 1 [T ] 2 [T ] 3 [T ] and
 get
c123 −s123 0 l1 c1 + l2 c12

0 [T ] =  s123 c123 0 l1 s1 + l2 s12 

3  0 0 1 0 
0 0 0 1
To obtain 0 0 3
Tool [T ], multiply 3 [T ] Tool [T ]
 
c123 −s123 0 l1 c1 + l2 c12 + l3 c123
 
0  s123 c123 0 l1 s1 + l2 s12 + l3 s123 
Tool [T ] =  0 0 1 0 
0 0 0 1
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PUMA 560 M ANIPULATOR
The PUMA 560 is a six-degree-of-freedom manipulator with all rotary joints —
Figure shows assigned coordinate systems.
Ẑ2
{2}
X̂2 Ẑ1
Ŷ2 {1}
Ŷ1 Ẑ3
{3}
O3
X̂1
a2 X̂3
Ŷ3
d4
X̂4
d3
X̂3 {4} Ẑ4
O1 , O2
{3}
a3
O4
Ŷ3 X̂6
O3 X̂5
Ŷ4 X̂4 {6} {5}
Ẑ6 Ŷ5
{4}
Ẑ4 O4 , O5 , O6
(a) The PUMA 560 manipulator (b) PUMA 560 - forearm and wrist
. . . . . . . . . . . . . . . . . . . .
F IGURE : The PUMA 560 manipulator . . . . . . . . . . . . . . . . . . . .
PUMA 560 M ANIPULATOR – D-H
PARAMETERS
The D-H parameters for the PUMA 560 manipulator are
1 0 0 0 θ1
2 −π /2 0 0 θ2
3 0 a2 d3 θ3
4 −π /2 a3 d4 θ4
5 π /2 0 0 θ5
6 −π /2 0 0 θ6
Note: θi , i = 1, 2, ..., 6 are the six joint variables.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PUMA 560 M ANIPULATOR – L INK
TRANSFORMS
Substituting
 elements of row #1 of D-H table and using equation (10), we get
c1 −s1 0 0

0 [T ] =  s1 c1 0 0  
1  0 0 1 0 
0 0 0 1
 
c2 −s2 0 0
 0 0 1 0 
From row # 2, we get 12 [T ] = 
 −s2 −c2 0 0 

0 0 0 1
 
c3 −s3 0 a2
 s 3 c3 0 0 
From row # 3, we get 23 [T ] = 
 0

0 1 d3 
0 0 0 1 . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PUMA 560 M ANIPULATOR : L INK
 
c4 −s4 0 a3
 0 0 1 d4 
From row # 4 , we get 34 [T ] = 
 −s4 −c4 0

0 
0 0 0 1
 
c5 −s5 0 0
 0 0 −1 0 
From row # 5 , we get 45 [T ] = 
 s5 c5

0 0 
0 0 0 1
 
c6 −s6 0 0
 0 0 1 0 
From row # 6 , we get 56 [T ] = 
 −s6 −c6 0

0 
0 0 0 1
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PUMA 560 M ANIPULATOR : L INK
{3} with respect to {0}is given by 

c1 c23 −c1 s23 −s1 a2 c1 c2 − d3 s1

0 [T ] = 0 [T ]1 [T ]2 [T ] =  s1 c23 −s1 s23 c1 a2 s1 c2 + d3 c1  
3 1 2 3  −s23 −c23 0 −a2 s2 
0 0 0 1
{6} with respect to {3}is given by 
c4 c5 c6 − s4 s6 −c4 c5 s6 − s4 c6 −c4 s5 a3

3 [T ] = 3 [T ]4 [T ]5 [T ] =  s5 c6 −s5 s6 c5 d4 

6 4 5 6  −s4 c5 c6 − c4 s6 s4 c5 s6 − c4 c6 s4 s5 0 
0 0 0 1
Can obtain any required link transformation matrix once D-H is table known!
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
SCARA M ANIPULATOR
Ẑ 2
Ẑ 1 θ2
{ 2}
θ1
{ 1}
SCARA – Selective Compliance
Assembly Robot Arm
X̂ 2
X̂ 1 Very popular for robotic
d3
{ 3}
assembly
O 0, O1 O2 Ẑ 3 { 4}
Capability of desired compliance
O3, O4 X and rigidity in selected directions.
Three rotary(R) joint and one
θ4
prismatic (P) joint.
A 4 DOF manipulator.
F IGURE : A SCARA manipulator

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
SCARA M ANIPULATOR : D-H
PARAMETERS
{0} and {1} have same origin & origins of {3} and {4} chosen at the base of
the parallel jaw gripper.
Directions of Ẑ3 chosen pointing upward (see Figure).
Note: Actual SCARA manipulator has ball-screw at the third joint — We
assume P joint.
D-H Table for SCARA robot
1 0 0 0 θ1
2 0 a1 0 θ2
3 0 a2 −d3 0
4 0 0 0 θ4
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
SCARA M ANIPULATOR – L INK
TRANSFORMS
Using equation
 (10) and the  D-H table,link transforms can 
be obtained as
c1 −s1 0 0 c2 −s2 0 a1
 c1 0 0   
0 [T ] =  s1  , 1 [T ] =  s2 c2 0 0 
1  0 0 1 0  2  0 0 1 0 
0 0 0 1 0 0 0 1
   
1 0 0 a2 c4 −s4 0 0

2 [T ] =  0 1 0 0   s4 c4 0 0 
 3  
3  0 0 1 −d3  , 4 [T ] =  0 0 1 0 
0 0 0 1 0 0 0 1
The transformation matrix 4 [T 0
 ] is 
c124 −s124 0 a1 c1 + a2 c12

0 [T ] = 0 [T ]1 [T ]2 [T ] 3 [T ] =  s124 c124 0 a1 s1 + a2 s12 
4 1 2 3 4  0 0 1 −d3 
0 0 0 1 . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PARALLEL M ANIPULATORS
Extend idea of D-H parameters to a closed-loop mechanism/parallel

manipulator.
Key idea – Break a parallel manipulator into serial manipulators.
Obtain D-H parameters for serial manipulators.
Several ways to break
Choose one that leads to simple serial manipulators
During analysis – Combine serial manipulators using constraints
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PARALLEL M ANIPULATORS : 4- BAR
M ECHANISM
Link 2
l2 Rotary joints 1 and 4 fixed

to ground at two places.
φ2
Ŷ L
Link 1 Link 3 l3 ŶR Break 4-bar mechanism into
{L} l1
{R}
two serial manipulators.
φ1 Break at joint 3 → A 2R
θ1 X̂L
l0 planar manipulator + a 1R
OR X̂R manipulator
OL
F IGURE : A planar four-bar mechanism

. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
4- BAR M ECHANISM – D-H
PARAMETERS
For 2R planar manipulator – D-H parameter with respect to {L}

iαi−1 ai−1 di θi
1 0 0 0 θ1
2 0 l1 0 ϕ2
For 1R planar manipulator – D-H parameters with respect to {R}
1 0 0 0 ϕ1
The constant transform LR [T ] is known.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PARALLEL M ANIPULATORS : T HREE
DOF E XAMPLE
p(x, y, z) Moving Platform
Ẑ
Spatial 3- DOF parallel manipulator.
S3 S2
Top (moving) platform & (fixed)

bottom platform are equilateral
l2
S1
triangles.
l3 θ2
l1 Figure shows the home position and
Axis of R2
θ3 {Base} Ẑ need not pass through top
O
platform centre always.
Ŷ Three legs — Each leg is of R-P-S
θ1 configuration.
Base Platform
Axis of R3
Three P joints actuated.
X̂
Axis of R1 First proposed by Lee and Shah
(1988) as a parallel wrist – not a
wrist! . . . . . . . . . . . . . . . . . . . .
F IGURE : A three DOF parallel manipulator . . . . . . . . . . . . . . . . . . . .
T HREE DOF E XAMPLE : D-H
PARAMETERS
D-H parameter for first leg with respect to {L1 }

1 0 0 0 ϕ1
2 −π /2 0 l1 0
D-H parameter for all R-P-S leg same except the reference coordinate system.
{L1 }, {L2 }, and {L3 } are at the three rotary joints R1 , R2 , and R3 , respectively.
{Base} is located at the centre of the base platform & Base Li [T ], i = 1, 2, 3, are
constant and known.
Note: The angle θ1 shown in figure is same as π /2 − ϕ1 .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
T HREE DOF E XAMPLE : L INK
TRANSFORMATION MATRICES
   
c1 −s1 0 0 1 0 0 0
 s1 c1 0 0  
1 [T ] =  0 0 1 l1 
L1  , 
1 [T ] =  0 0 1 0  2  0 −1 0 0 
0 0 0 1 0 0 0 1
2 [T ]
S1 is an identity matrix – {S1 } is located at the centre of the spherical joint
and parallel to {2}.
Base [T ] L1 1 2
S1 = Base
L1 [T ]1 [T ]2 [T ]S1 [T ]
Location of spherical joint S1 with respect to {Base} from Base S1 [T ]
Base S
1 = (b − l 1 cos θ1 , 0, l1 sin θ1 )T , b is the distance of R from the origin of
1
{Base} (see figure).
√ √
2 + 2 l2 cos θ2 , 2 − 2 cos θ2 , l2 sin θ2 )T
Base S b 1 3b 3l2
Location of S2 and S3 2√= (−√
2 l3 cos θ3 , − 2 + 2 cos θ3 , l3 sin θ3 )

1 3b 3l3
Base S
3 = (− b2 + T
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
PARALLEL M ANIPULATORS : S IX
DOF E XAMPLE
Z
q2
d f2
q1 d Moving platform connected to
l 21 l23
b2 l22 fixed base by three chains.
y2 Each chain is R-R-R & S joint at
b1 l11 p2
l12 Y
l 13 s top.
h f1 s
y1 Model of a three-fingered hand
p1 s
q3 p3 (Salisbury, 1982) gripping an
l 33
object with point contact and
l 31 no-slip.
b3 l32 f3
Each finger modeled with R-R-R
y3
joints and point of contact
X modeled as S joint.
. . . . . . . . . . . . . . . . . . . .
F IGURE : A six DOF parallel (hybrid) manipulator . . . . . . . . . . . . . . . . . . . .
S IX DOF E XAMPLE : D-H
PARAMETERS AND LINK TRANS -
FORMS
D-H parameters for R-R-R chain.

i αi−1 ai−1 θi di
1 0 0 θ1 0
2 π /2 l11 ψ1 0
3 0 l12 ϕ1 0
D-H parameter does not contain last link length l13 .
D-H parameters for three fingers with respect to {Fi }, i = 1, 2, 3 identical.
Can obtain transformation matrix Fpii [T ] by matrix multiplication.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
S IX DOF
E XAMPLE : L INK
Positionvector of spherical joint i 

cos θi (li1 + li2 cos ψi + li3 cos(ψi + ϕi ))
Fi p =  sin θ (l + l cos ψ + l cos(ψ + ϕ )) 
i i i1 i2 i i3 i i
li2 sin ψi + li3 sin(ψi + ϕi )
With respect to {Base}, the locations of {Fi }, i = 1, 2, 3, are known and
constant (see Figure)
Base b = (0, −d, h)T Base b = (0, d, h)T Base b = (0, 0, 0)T
1 2 3
Orientation of {Fi }, i = 1, 2, 3, with respect to {Base} are also known - {F1 }
and {F2 } are parallel to {Base} and {F3 } is rotated by γ about the Ŷ (not
shown in figure!).
The transformation matrices Base [T ] is Base [T ]0 [T ]1 [T ]2 [T ]3 [T ] – Last
pi F1 1 2 3 p1
transformation includes l13 .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
S IX DOF
E XAMPLE : L INK
Extract the position vector Base p1 from the last column of Base F1 [T ]
 
cos θ1 (l11 + l12 cos ψ1 + l13 cos(ψ1 + ϕ1 ))
Base p =Base b +F1 p =  −d + sin θ (l + l cos ψ + l cos(ψ + ϕ )) 
1 1 1 1 11 12 1 13 1 1
h + l12 sin ψ1 + l13 sin(ψ1 + ϕ1 )
Similarly for
 second leg 
Base p =  d + sin θ (l + l cos ψ + l cos(ψ + ϕ )) 
2 2 21 22 2 23 2 2
h + l22 sin ψ2 + l23 sin(ψ2 + ϕ2 )
 
For third leg Base p3 = [R(Ŷ, γ )]  sin θ3 (l31 + l32 cos ψ3 + l33 cos(ψ3 + ϕ3 )) 
l32 sin ψ3 + l33 sin(ψ3 + ϕ3 )
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
S UMMARY
D-H parameters obtained for serial manipulators – Planar 3R, PUMA 560,
SCARA
To obtain D-H parameters for parallel manipulators
Break parallel manipulator into serial manipulators.
Obtain D-H parameters for each serial chain.
Examples of 4-bar mechanism, 3-degree-of-freedom and 6-degree-of-freedom
parallel manipulators.
Can extract position vectors of point of interest & orientation of links from link
transforms.
Kinematic analysis can be done using the concepts presented here.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
Suggested Additional Reading – Week -2
• Reference textbooks in orientation of rigid bodies, theoretical kinematics (screws, twists and
wrenches), D-H parameters conventions (original and modern), advanced mathematical treat-
ment of robotics.
– T. R. Kane, P. W. Likins, and D. A. Levinson 1983, Spacecraft Dynamics, McGraw-Hill

Inc.
– J. M. McCarthy 1990, An Introduction to Theoretical Kinematics, MIT Press.
– J. J. Craig 1989, Introduction to Robotics: Mechanics and Control, 2nd Ed. Addison-
Wesley.
– R. S. Hartenberg and J. Denavit 1964, Kinematic Synthesis of Linkages, McGraw-Hill
Book Company.
– R. M. Murray, Z. Li and S. S. Sastry 1994, A Mathematical Introduction to Robotic
Manipulation, CRC Press.
• K. M. Lee and D. Shah 1988, “Kinematic analysis of a three-degrees-of-freedom in-parallel

actuated manipulator”, IEEE Trans. on Robotics and Automation, Vol. 14, No. 4, pp.
354-360.
• R. Sangamesh Deepak and A. Ghosal 2006, “A note on the diagonalizability and the Jor-
dan form of the 4 × 4 homogeneous transformation matrix”, Trans. of ASME, Journal of
Mechanical Design, Vol. 128, No. 6, pp. 1343-1348.
• G. Arfken 1985, Mathematical Methods for Physicists, 3rd Ed., Academic Press.
• About quaternions – http://mathworld.wolfram.com/Quaternion.html

Lecture Notes W2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture Notes W2

Uploaded by

Copyright:

Available Formats

ROBOTICS : BASIC AND A DVANCED C ONCEPTS

M ATHEMATICAL P RELIMINARIES , D-H C ONVENTION & E XAMPLES

Department of Mechanical Engineering

Point A p with Cartesian coordinates (px , py , pz )T

F IGURE : Orientation of a rigid body . . . . . . . . . . . . . . . . . . . .

rij , i, j = 1, 2, 3 are called direction cosines

The vector B p in rigid body B can be described in {A} by

B [R] in terms of (kx , ky , kz ) and angle ϕ

r11 = kx 2 (1 − cos ϕ ) + cos ϕ

F IGURE : Rotation about X̂ by ϕ . . . . . . . . . . . . . . . . . . . .

Rotation about Ŷ and Ẑ

Orientation described by 3 independent parameters → Three rotations

ŶA ŶA θ2 ŶA

F IGURE : X–Y–Z Euler angles . . . . . . . . . . . . . . . . . . . .

X̂B1 X̂B1 X̂B2 X̂B1 X̂B

F IGURE : Z–Y–Z Euler angles

Euler parameters (see Kane et al., 1983, McCarthy, 1990): 4 parameters

Orientation of a rigid body in ℜ3 is specified by 3 independent parameters.

F IGURE : General transformation

Combined translation and orientation

Four eigenvalues of A ±ıϕ , ϕ same as for A [R].

A joint connects two or more links.

F IGURE : – Types of joints . . . . . . . . . . . . . . . . . . . .

ROTARY (R) JOINT Rigid Body i − 1 θi

Rigid bodies, {i − 1} and {i}, Oi

connected by a rotary (R) joint.

{i} can rotate about k̂, with respect Oi−1

matrix equation above; θi is

unknown⇒ 2 constraints in the 3 {0}

F IGURE : A rotary joint

Rotary joint in a loop — 2 Rigid Body i

ends {L} and {R}.

F IGURE : A rotary joint in a loop

Spherical(S) or ball and socket

S joint can be represented as 3 0

intersecting rotary(R) joints.

F IGURE : A spherical joint

The S–S pair appear in

Distance between two

F IGURE : A S–S pair in a loop

A link A is a rigid body in 3D space → Can be described by a coordinate

A word about conventions – Several ways to derive D-H parameters!

Convention same as in Craig (1986) or Ghosal(2006).

For coordinate system {i − 1}

Location of origin of {i} with respect to {i − 1} is given by

Steps to obtain D-H parameters for a n link serial manipulator

All 3 rotary joint axis Ẑi parallel O3

and pointing out. {3}

top, respectively. {0}

Origin O0 is coincident with O1 {1}

F IGURE : The planar. 3R manipulator

The D-H parameters for the PUMA 560 manipulator are

Note: θi , i = 1, 2, ..., 6 are the six joint variables.

{3} with respect to {0}is given by 

F IGURE : A SCARA manipulator

Extend idea of D-H parameters to a closed-loop mechanism/parallel

l2 Rotary joints 1 and 4 fixed

F IGURE : A planar four-bar mechanism

For 2R planar manipulator – D-H parameter with respect to {L}

Top (moving) platform & (fixed)

D-H parameter for first leg with respect to {L1 }

2 l3 cos θ3 , − 2 + 2 cos θ3 , l3 sin θ3 )

D-H parameters for R-R-R chain.