Robotics 1
Direct kinematics
Prof. Alessandro De Luca
Robotics 1
Kinematics of robot manipulators
!
study of ...
geometric and timing aspects of robot motion,
without reference to the causes producing it
robot seen as ...
an (open) kinematic chain of rigid bodies
interconnected by (revolute or prismatic) joints
Robotics 1
Motivations
!
functional aspects
!
!
definition of robot workspace
calibration
operational aspects
task execution
(actuation by motors)
task definition and
performance
two different spaces related by kinematic (and dynamic) maps
!
!
!
Robotics 1
trajectory planning
programming
motion control
3
Kinematics
formulation and parameterizations
r = f(q)
JOINT
space
q = (q1,,qn)
!
q = f -1(r)
r = (r1,,rm)
choice of parameterization q
!
!
INVERSE
TASK
(Cartesian)
space
DIRECT
unambiguous and minimal characterization of robot configuration
n = # degrees of freedom (dof) = # robot joints (rotational or
translational)
choice of parameterization r
!
compact description of position and/or orientation (pose) variables of
interest to the required task
usually, m ! n and m ! 6 (but none of these is strictly necessary)
Robotics 1
Open kinematic chains
q2
q4
q1
q3
e.g., the relative angle
between a link and the
following one
!
r = (r1,,rm)
e.g., it describes the
pose of frame RFE
pointing in space
positioning in the plane
m=3
!
!
Robotics 1
RFE
m=2
!
qn
orientation in space
positioning and orientation in the plane
5
Classification by kinematic type
(first 3 dofs)
SCARA
(RRP)
Cartesian or
gantry
(PPP)
cylindric
(RPP)
articulated or
anthropomorphic
(RRR)
polar or
spherical
(RRP)
Robotics 1
R = 1-dof rotational (revolute) joint
P = 1-dof translational (prismatic) joint
Direct kinematic map
!
the structure of the direct kinematics function
depends from the chosen r
r = fr(q)
!
methods for computing fr(q)
!
!
Robotics 1
geometric/by inspection
systematic: assigning frames attached to the robot
links and using homogeneous transformation matrices
Example: direct kinematics of 2R arm
y
py
l2
l1
"
q2
q1
px
q1
q=
q2
n=2
px
r = py
"
m=3
px = l1 cos q1 + l2 cos(q1+q2)
py = l1 sin q1 + l2 sin(q1+q2)
" = q1+ q2
for more general cases, we need a method!
Robotics 1
Numbering links and joints
joint 2
joint i+1
link i-1
link 1
joint 1
link 0
(base)
joint i-1
revolute
Robotics 1
link i
joint i
joint n
link n
(end effector)
prismatic
9
Spatial relation between joint axes
axis of joint i+1
axis of joint i
90
A
%$
# i$
90
B
common normal
(axis of link i)
a i = displacement AB between joint axes (always well defined)
# i = twist angle between joint axes
projected on a plane % orthogonal to the link axis
Robotics 1
with sign
(pos/neg)!
10
Spatial relation between link axes
axis of joint i
link i-1
link i
D
C
axis of link i-1
axis of link i
&i
'$
d i = displacement CD (a variable if joint i is prismatic)
& i = angle between link axes (a variable if joint i is revolute)
projected on a plane ' orthogonal to the joint axis
Robotics 1
with sign
(pos/neg)!
11
Denavit-Hartenberg (DH) frames
joint axis
i-1
joint axis
i
link i-1
link i
link i-1 is moved
by joint i-1
frame RFi is
attached to link i
zi-1
xi-1
di
Oi-1
axis around which the link rotates
or along which the link slides
Robotics 1
joint axis
i+1
ai
zi
Oi
&i
#i
xi
common normal
to joint axes
i and i+1
12
Denavit-Hartenberg parameters
axis of joint
i
axis of joint
i-1
link i
link i-1
D
zi-1
xi-1
di
Oi-1
!
!
!
!
!
!
ai
axis of joint
i+1
zi
Oi
&i
#i
xi
unit vector zi along axis of joint i+1
unit vector xi along the common normal to joint i and i+1 axes (i ( i+1)
ai = distance DOi positive if oriented as xi (constant = length of link i)
di = distance Oi-1D positive if oriented as zi-1 (variable if joint i is PRISMATIC)
#i = twist angle between zi-1 and zi around xi (constant)
&i = angle between xi-1 and xi around zi-i (variable if joint i is REVOLUTE)
Robotics 1
13
Denavit-Hartenberg layout made simple
(a popular 3-minute illustration...)
video
https://www.youtube.com/watch?v=rA9tm0gTln8
!
note: the authors of this video use r in place of a, and do not add subscripts!
Robotics 1
14
Ambiguities in defining DH frames
!
!
Robotics 1
frame0: origin and x0 axis are arbitrary
framen: zn axis is not specified (but xn must be
orthogonal to and intersect zn-1)
when zi-1 and zi are parallel: the common
normal is not uniquely defined (Oi can be
chosen arbitrarily along zi)
when zi-1 and zi are incident: the positive
direction of xi can be chosen at will (however,
we often take xi = zi-1 ! zi)
15
Homogeneous transformation
between successive DH frames (from framei-1 to framei)
!
roto-translation around and along zi-1
c&i -s&i 0 0
s&i c&i 0 0
i-1A (q ) =
i
i
0 0 1 0
0 0
0 1
1 0 0 0
0 1 0 0
0 0 1 di =
0 0 0 1
rotational joint ) qi = &i
!
c&i -s&i 0 0
s&i c&i 0 0
0 0 1 di
0 0
0 1
prismatic joint ) qi = di
roto-translation around and along xi
1 0 0 ai
0 c#i -s#i 0
iA =
i
0 s#i c#i 0
Robotics 1
0 0
always a
constant matrix
16
Denavit-Hartenberg matrix
J. Denavit and R.S. Hartenberg, A kinematic notation for lower-pair mechanisms based on matrices,
Trans. ASME J. Applied Mechanics, 23: 215221, 1955
c&i
s&i
i-1A (q ) = i-1A (q ) iA =
i
i
i
i
i
0
0
-c#i s&i
c#i c&i
s#i
s#i s&i
-s#i c&i
c#i
aic&i
ais&i
di
compact notation: c = cos, s = sin
super-compact notation: ci = cos qi, si = sin qi
Robotics 1
17
Direct kinematics of manipulators
yE
RFE
description internal
to the robot using
product 0A1(q1) 1A2(q2)n-1An(qn)
q = (q1,,qn)
slide s
zE
xE
z0
approach a
normal n
RF0
RFB
y0
x0
description external
to the robot using
BTE
BT
E
= BT0 0A1(q1) 1A2(q2) n-1An(qn) nTE
R p
000 1
nsa p
000 1
r = (r1,,rm)
r = fr(q)
Robotics 1
alternative descriptions of the direct kinematics of the robot
18
Example: SCARA robot
video
q3
q1
q2
q4
Sankyo SCARA 8438
Robotics 1
Sankyo SCARA SR 8447
19
Step 1: joint axes
all parallel
(or coincident)
twists # i = 0
or %
J2 elbow
J3 prismatic
J4 revolute
J1 shoulder
Robotics 1
20
Step 2: link axes
the vertical heights
of the link axes
are arbitrary
(for the time being)
a1
Robotics 1
a2
a3 = 0
21
Step 3: frames
z1
x1
axes yi for i > 0
are not shown
z2 = z3
x2
(nor needed; they form
right-handed frames)
x4
z0
x3
z4 = a axis
(approach)
x0
Robotics 1
y0
22
Step 4: DH table of parameters
z1
x1
z2 = z3
x2
x3
x4
z0
x0
Robotics 1
z4
y0
# i$
ai
di
&i
a1
d1
q1
a2
q2
q3
d4
q4
note that:
d1 and d4 could be set = 0
here, it is d4 < 0
23
Step 5: transformation matrices
c&1 - s&1
c&1
0A (q ) = s&1
1 1
0
0
0
0
0 a1c&1
0 a1s&1
1
d1
0
1
c&2 - s&2
c&2
1A (q ) = s&2
2 2
0
0
0
0
0 a2c&2
0 a2s&2
1
0
0
1
q = (q1, q2, q3, q4)
= (&1, &2, d3, &4)
Robotics 1
1
2A (q ) = 0
3 3
0
0
c&4
3A (q ) = s&4
4 4
0
0
0
1
0
0
0
0
1
0
0
0
d3
1
s&4
-c&4
0
0
0
0
-1
0
0
0
d4
1
24
Step 6a: direct kinematics
as homogeneous matrix BTE (products of iAi+1)
c12 -s12
0A (q ,q ,q ) = s12 c12
3 1 2 3
0
0
0
0
0
0
1
0
c4
3A (q ) = s4
4 4
0
0
0
0
-1
0
R(q1,q2,q4)=[ n s a ]
s4
-c4
0
0
c124 s124 0
0
BT = 0A (q ,q ,q ,q ) = s124 -c124
E
4 1 2 3 4
0
0
-1
(BT0 = 4TE = I)
0
0
0
Robotics 1
a1c1+ a2c12
a1s1+ a2s12
d1+q3
1
0
0
d4
1
p = p(q1,q2,q3)
a1c1+ a2c12
a1s1+ a2s12
d1+q3+d4
1
25
Step 6b: direct kinematics
as task vector r IRm
c124 s124 0
0
0A (q ,q ,q ,q ) = s124 -c124
4 1 2 3 4
0
0
-1
0
0
0
extract #z
from
R(q1,q2,q4)
Robotics 1
a1c1+ a2c12
a1s1+ a2s12
d1+q3+d4
1
px
a1c1+ a2c12
r = py = fr(q) = a1s1+ a2s12 IR4
pz
d1+q3+d4
#z
q1+q2+q4
take
p(q1,q2,q3)
as such
26
Stanford manipulator
!
6-dof: 2R-1P-3R (spherical wrist)
!
!
shoulder offset
one possible DH assignment of
frames is shown
determine the associated
!
!
write a program for computing
the direct kinematics
!
!
Robotics 1
DH parameters table
homogeneous transformation
matrices
direct kinematics
numerically (Matlab)
symbolically (Mathematica,
Maple, Symbolic Manipulation
Toolbox of Matlab, )
27
DH table for Stanford manipulator
!
6-dof: 2R-1P-3R (spherical wrist)
ai
di
-!/2
d1>0
q1=0
!/2
d2>0
q2=0
q3>0
-!/2
-!/2
q4=0
!/2
q5=-!/2
d6>0
q6=0
joint variables are in red,
with their current value in the shown configuration
Robotics 1
28
KUKA KR5 Sixx R650
!
6R (offsets at shoulder and elbow, spherical wrist)
top view
!
determine
!
!
!
Robotics 1
side view (from an observer in V)
frames and table of DH parameters
homogeneous transformation matrices
direct kinematics
available at
DIAG Robotics Lab
29
KUKA LWR 4+
!
7R (no offsets, spherical shoulder and spherical wrist)
available at
DIAG Robotics Lab
!
determine
!
!
!
front
Robotics 1
front view
frames and table
of DH parameters
homogeneous
transformation
matrices
direct kinematics
d1 and d7 can be
set = 0 or not
(as needed)
back
side view
30
Appendix:
Modified DH convention
!
a modified version used in J. Craigs book Introduction to Robotics, 1986
!
!
!
!
!
has zi axis on joint i
ai & #i = distance & twist angle from zi to zi+1, measured along & about xi
di & &i = distance & angle from xi-1 to xi, measured along & about zi
source of much confusion... if you are not aware of it (or dont mention it!)
convenient with link flexibility: a rigid frame at the base, another at the tip...
y2
classical % c# i
i"1
' s# i
Ai = '
0
'
& 0
"c$ i s# i
c$ i c# i
s$ i
0
s$ i s# i
"s$ i c# i
c$ i
0
ai c# i (
ai s# i *
di *
*
1 )
planar 2R
example
y1
y0
x2
x1
x0
modified %
i"1
Aimod
c# i
'c$ i"1 s# i
='
s$ s#
' i"1 i
& 0
Robotics 1
"s# i
c$ i"1 c# i
s$ i"1 c# i
0
0
ai"1 (
"s$ i"1 "di s$ i"1 *
c$ i"1
di c$ i"1 *
*
0
1 )
y1
y0
y2
x2
x1
x0
modified DH tends to place frames
at the base of each link
31