1

@ McGraw-Hill Education

Lecture 3
Robot Kinematics
by
S.K. Saha
Aug. 22’16 (M)@JRL301 (Robotics Tech.)

PROPRIETARY MATERIAL. © 2014, 2008 The McGraw-Hill Companies, Inc. All rights reserved. No part of this PowerPoint slide may be displayed,
reproduced or distributed in any form or by any means, without the prior written permission of the publisher, or used beyond the limited distribution to teachers
and educators permitted by McGraw-Hill for their individual course preparation. If you are a student using this PowerPoint slide, you are using it without
permission.
2
@ McGraw-Hill Education

Recap

• Kinematic chain: Links and joints

• DOF: Parameters-constraints
• Position: Simple (like good friend in the
hostel)
• Orientation: Confusing and SERIOUS
attention to be paid
3
@ McGraw-Hill Education

Denavit and Hartenberg (DH)

Parameters—Frame Allotment

• Serial chain
- Two links connected
by revolute joint, or
- Two links connected
by prismatic joint

Fig. 5.27
4
@ McGraw-Hill Education

• Joint axis i: Link i-1 + link i

• Link i: Fixed to frame i+1 (Saha) / frame i (Craig)
DH Variables Constants
bi and θi ai and αi
[Screw@Z] [Screw@X]
Saha XiXi+1@Zi ZiZi+1@Xi+1 Zi+1

Craig Xi-1Xi@Zi ZiZi+1@Xi

Z’’’i
5
@ McGraw-Hill Education

Revolute Joint
• DH@Z (Variable) • DH@X (Const.)
– Joint offset (b) – Link length (a)
– Joint angle (θ) – Twist angle (α)
Zi+1

Z’’’i

Fig. 5.28
6
@ McGraw-Hill Education

Mathematically
• Translation along Zi
⎡1 0 0 0⎤
⎢0 1 0 0 ⎥⎥
Tb = ⎢ . . . (5.49a)
⎢0 0 1 bi ⎥
⎢ ⎥
⎣0 0 0 1⎦

• Rotation about Zi
⎡C θ i − S θi 0 0⎤
⎢ Sθ Cθi 0 0 ⎥⎥ . . . (5.49b)
Tθ = ⎢ i
⎢ 0 0 1 0⎥
⎢ ⎥
⎣ 0 0 0 1⎦
7
@ McGraw-Hill Education

• Translation along Xi+1

⎡1 0 0 ai ⎤
⎢0 1 0 0 ⎥⎥
Ta = ⎢ . . . (5.49c)
⎢0 0 1 0⎥
⎢ ⎥
⎣0 0 0 1⎦

• Rotation about Xi+1

⎡1 0 0 0⎤
⎢0 C αi − S αi 0 ⎥⎥
Tα = ⎢ . . . (5.49d)
⎢0 Sα i Cαi 0⎥
⎢ ⎥
⎣0 0 0 1⎦
8
@ McGraw-Hill Education

• Total transformation from Frame i to Frame i+1

Ti = TbTθTaTα . . . (5.50a)

⎡C θ i − Sθ i C α i Sθ i Sα i aiCθ i ⎤

Position
⎢ Sθ Rotation
Cθ Cα − C θ Sα a i Sθ i ⎥⎥
Ti = ⎢ i i i i i
⎢ 0 Matrix
Sα iCα i bi ⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦

. . . (5.50b)
9
@ McGraw-Hill Education

Three-link Planar Arm

• DH-parameters
Link bi θi ai αi
1 0 θ1 (JV) a1 0
2 0 θFill-up
(JV)the DH
2parametersa2 0
3 0 θ3 (JV) a3 0

• Frame transformations
(Homogeneous)
⎡ C θ i − Sθ i 0 a i C θ i ⎤
⎢ SθFill-upCwith
θi
the elements ⎥
0 a i Sθ i ⎥ ,
Ti = ⎢ i for i=1,2,3
⎢ 0 0 1 0 ⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦
10
@ McGraw-Hill Education

Forward and Inverse Kinematics

Multiply
+ Add
Forward: One soln.

Non-lin. eqns.
Inverse: 1st soln.

Solve
.
Inverse: nth soln.
11
@ McGraw-Hill Education

Overall Transformation for Three-link Arm

T = T1 T2 T3
⎡ C θ 123 − Sθ 123 0 a 1C θ 1 + a 2 C θ 12 + a 3C θ 123 ⎤
⎢ Sθ C θ 123 0 a 1 S θ 1 + a 2 S θ 12 + a 3 S θ 123 ⎥
T= ⎢ 123 ⎥
⎢ 0 0 1 0 ⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦
12
@ McGraw-Hill Education

DH Parameters of Articulated Arm

Link bi θi ai αi
1 0 θ1 (JV) 0 − π/2
2 0 θ2 (JV) a2 0
3 0 θ3 (JV) a3 0
13
@ McGraw-Hill Education

Matrices for Articulated Arm

⎡c1 0 − s1 0⎤ ⎡ c2 − s2 0 a2 c2 ⎤
⎢s 0 c 0 ⎥⎥ ⎢s c2 0 a 2 s 2 ⎥⎥
T1 = ⎢ 1 1
T2 ≡ ⎢ 2
⎢ 0 −1 0 0⎥ ⎢0 0 1 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣0 0 0 1⎦ ⎣0 0 0 1 ⎦
⎡ c3 − s3 0 a 3 c3 ⎤
⎢s c3 0 a 3 s 3 ⎥⎥
T3 ≡ ⎢ 3
⎢0 0 1 0 ⎥
⎢ ⎥
⎣0 0 0 1 ⎦

⎡ c 1 c 23 - c 1 s 23 − s1 c 1 (a 2 c 2 + a 3 c 23 ) ⎤
⎢s c − s 1 s 23 c1 s 1 (a 2 c 2 + a 3 c 23 ) ⎥⎥
T ≡ ⎢ 1 23
… (6.11)
⎢ − s 23 − c 23 0 − ( a 2 s 2 + a 3 s 23 ) ⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦
14
@ McGraw-Hill Education

Inverse Kinematics
• Unlike Forward Kinematics, general solutions
are not possible.

• Several architectures are to be solved

differently.
15
@ McGraw-Hill Education

Two-link Arm
Y1 X3

RoboAnalyzer
p x = a1c1 + a2 c12 Y3
py
p y = a1s1 + a2 s12 θ2
a2 θ X
2 2
p x2 + p y2 − a12 − a 22 Y2
c2 =
2 a1 a 2 a1
s 2 = ± 1 − c 22 θ1
θ1
θ2 = atan2 (s2, c2) px X1
(a 1 + a 2 c 2 ) p y − a 2 s 2 p x
s1 = Δ ≡ a 2
+ a 2
+ 2 a a c = p 2
+ p 2
Δ 1 2 1 2 2 x y
(a 1 + a 2 c 2 ) p x + a 2 s 2 p y
c1 = θ1 = atan2 (s1, c1)
Δ
16
@ McGraw-Hill Education

Inverse Kinematics of 3-DOF RRR Arm

φ = θ1 + θ 2 + θ3 … (6.18a)
p x = a1 c1 + a 2 c12 + a 3 c123
… (6.18b)
p y = a1 s1 + a 2 s12 + a 3 s123
… (6.18c)

w x = p x − a 3 c φ = a1 c1 + a 2 c12 … (6.19a)
w y = p y − a 3 s φ = a1 s1 + a 2 s12 … (6.19b)
17
@ McGraw-Hill Education

w2x + w2y = a12+ a22 + 2 a1a2c2 … (6.20a)

w 12 + w 22 − a 12 − a 22
c2 = s 2 = ± 1 − c 22 … (6.20b,c)
2 a1 a 2
θ2 = atan2 (s2, c2) . . . (6.21)
wx = ( a1 + a2 c2 )c1 − a2 s1s2 … (6.22a)
w y = (a1 + a 2 c 2 )s1 + a 2 c1 s 2 … (6.22b)
(a 1 + a 2 c 2 ) w y − a 2 s 2 w x (a 1 + a 2 c 2 ) w x + a 2 s 2 w y
s1 = c1 =
Δ Δ
Δ ≡ a12 + a 22 + 2 a1 a 2 c 2 = w x2 + w y2
… (6.23a,b)

θ1 = atan2 (s1, c1) . . . (6.23c)

θ3 = ϕ - θ1 − θ2 . . . (6.24)
18
@ McGraw-Hill Education

Numerical Example

• An RRR planar arm (Example 6.15). Input

⎡ 1 3 5⎤
⎢ Rotation0
2
− Origin
3+ ⎥
of end-
4.23
⎢ 2 2⎥
⎢
T ≡⎢
Matrix 0
3 1 effector
3
+1⎥
⎥ 1.86
2 frame
⎢ 2 2 ⎥ 0
⎢ 0 0 1 0 ⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦
where φ = 60o, and a1 = a2 = 2 units, and a3 = 1 unit.

Do it yourself Æ Verify using RoboAnalyzer

19
@ McGraw-Hill Education

Using eqs. (6.13b-c), c2 = 0.866, and s2 = 0.5,

θ2 = 30o

Next, from eqs. (6.16a-b), s1 = 0, and c1= 0.866.

θ1 = 0o.

Finally, from eq. (6.17) ,

θ3 = 30o.

Therefore θ1 = 0o θ2 = 30o, and θ3 = 30 …(6.30b)

The positive values of s2 was used in evaluating θ2 = 30o.

The use of negative value would result in :

θ1 = 30o θ2 = -30o, and θ3 = 60o …(6.30c)

20
@ McGraw-Hill Education

Extra Reading: Watch

• Forward and Inverse Kinematics: Watch 3/3 of
IGNOU Lectures [29min]
https://www.youtube.com/watch?v=duKD8cvtBTI
• For more clarity: Watch 12 of Addis Ababa
Lectures [77 min]
[https://www.youtube.com/watch?v=NXWzk1toze4
• Robotics (13 of Addis Ababa Lectures): Inverse
Kinematics [82 min]
https://www.youtube.com/watch?v=ulP3YiJLiEM
21
@ McGraw-Hill Education

Velocity Analysis
Jacobian maps joint rates into end-effector’s velocities. It
depends on the manipulator configuration.
⎡θ1 ⎤
⎡ω e ⎤ ⎢ ⎥
twistof end - effector : t e ≡ ⎢ ⎥ ; Joint rates : θ = ⎢ # ⎥
⎣ ve ⎦ ⎢θn ⎥
⎣ ⎦

t e = Jθ where J = [ j1 j2 " jn ] and

⎡ e1 e2 " en ⎤
J=⎢
⎣ e1 × a 1e e 2 × a 2e " e n × a ne ⎥⎦ . . (6.86)

⎡ ei ⎤ ⎡ 0 ⎤
ji ≡ ⎢ ⎥ , if Joint i is revolute ji ≡ ⎢ ⎥ , if Joint i is prismatic
⎣ei × aie ⎦ ⎣e i × a ie ⎦
22
@ McGraw-Hill Education

Jacobian of a 2-link Planar Arm

J = [e1 × a 1e e 2 × a 2e ]

where e1 ≡ e 2 ≡ [0 0 1]T

a1e ≡ a1 + a 2
≡ [a1c1 + a2 c12 a1s1 + a2 s12 0]T
a2e ≡ a2
≡ [a2 c12 a2 s12 0]T

⎡ − a1 s1 − a2 s12 − a2 s12 ⎤
Hence, J = ⎢
⎣ a1 c1 + a2 c12 a2 c12 ⎥⎦
23
@ McGraw-Hill Education

Example: Singularity of 2-link RR Arm

⎡− a1 s1 − a 2 s12 − a 2 s12 ⎤
J≡⎢ ⎥ θ2 = 0 or π
⎣ a1 c1 + a 2 c12 a 2 c12 ⎦
24
@ McGraw-Hill Education

Summary

• Forward Kinematics
• Inverse kinematics
– A spatial 6-DOF wrist-portioned has 8
solutions
• Velocity and Jacobian
25
@ McGraw-Hill Education

THANK YOU
saha@mech.iitd.ac.in
http://sksaha.com