Scara
Scara
Scara
Elsevier
Abstract
Padhy, S.K., On the dynamics of SCARA robot, Robotics and Autonomous Systems, 10 (1992) 71-78.
Assembly automation using robots has proven to be very successful. It has been shown that use of robots improves the
accuracy of assembly, and saves assembly time and cost as well. SCARA (Selective Compliance Assembly Robot Arm) type
robots have been widely used in industry for assembly operations in both the mechanical and electronics domains of
manufacturing. One of the important issues in designing an optimal manipulator is the dynamic behavior of the manipulator
which is highly non-linear and strongly coupled due to the interaction of inertial, centripetal, coriolis and gravitational
forces. Although extensive study involving optimal time study, dynamic study using the Lagrange method, path planning etc.
have been made, no closed solution for the dynamics of this important robot has been reported.
This paper presents the study of kinematics and dynamics of the SCARA assembly robot. Closed solutions for the
dynamics of this robot are derived. To study the variations of the torques of the links a computer code is developed. It is
found that the torque is independent of the angular position, which makes the robot highly compliant.
Oi the joint angle from Xi_ 1 axis to the X i This robot has high compliance in the x - y
axis about the Zi_ ~ axis (using the right plane in the sense that the arm moves freely to
hand rule) accomplish the assembly task accurately. A robot
Ri-i a 3 × 3 rotation matrix which transforms is said to be of high compliance if the arm moves
any vector with reference to coordinate a lot in response to a small force and the manipu-
frame (Xi, Y~, Z~) to the coordinate sys- lator is then said to be spongy or springy. On the
tem (Xi_l, Y-I' Zi-l) contrary, if it moves a little the compliance is low
Si sine 0~ and the manipulator is said to be stiff.
Sijk sine Oiik = sine{(0i + 0j) + Ok} Fig. 1 illustrates this 2R-P-R robot with 4 axes.
v, linear velocity of the coordinate system The first two axes are actuated by DC servo
(Xi, Y~, Z i) with respect to base coordi- motors through harmonic drives, whose backlash
nate system ( X 0, II0, Z0) is almost zero. These two axes allow the robot to
¢.Oi angular velocity of the coordinate sys- move in a plane parallel to the x-y plane within
tem (X/, Y/, Z i) with respect to base co- the work-space. The z-axis movement is made
ordinate system ( X 0, Yo, Zo) possible by the ball screw which converts the
rotational motion of the motor to translational
motion of the axis. The end effector is rotated by
1. I n t r o d u c t i o n a DC servo motor through a gear box.
d2
m
d3
t 2 ~ 3 x 3
I
z4
I
I I
effects such as blacklash were ignored. Develop- inverse kinematics of the SCARA robot is dis-
ment of an integrated simulation package has cussed.
been reported from Gold Star Company of South The translational and rotational relationship
Korea, by Lee et al. [2]. The fourth order between the links of this robot are described by
Rungga-Kutta method is employed as the inte- attaching coordinate frames according to the De-
gration technique. The approach is based on the navit-Hartenberg (D-H) notation. The expression
state equations that are integrated to find the for the end effector frame relative to the base
velocity and positions. The acceleration is com- frame is given by the ann matrix (T) as:
puted as a function of motor torque. Callaja et al. T =A ° --A°AI A2A3
--.. 1--12z13~,4,
[3] studied this robot to design the servos for it.
Their study was primarily focussed on the control where
aspects of this robot. The dynamics of this robot
C1 -S I 0 liCl
is studied by Ibrahim et al. [4] using polynomial
and NC2 trajectories. Their work included a soft- Ao = $1 C~ 0 liSt
ware development in FORTRAN 77 to study the 0 0 1 0
[ci
dynamic behavior with links subjected to different 0 0 0 1
velocity trajectories. They used the Lagrangian -S 2 0 12C2 ]
model taking advantage of recursive formulation.
C2 0 12S2/
However, there is no closed solution reported for
the dynamics of this robot. The dynamic study 0
can provide both qualitative and quantitative 0 0
analysis of this robot as well as can guide for a 0 0
better design of this robot. 1 0 0
3. Present work
0
0
1
0
-d 3
°1
1
C~ -$4 0 01
The present analysis of this robot is carried A3= $4 C4 0 0
out to study the dynamic behavior of this robot. 0 1 0
The kinematics is considered first to find the 0 0 1
relationship between the angular displacement
and the position of the end effector, and to Cl24 -5124 0 liC 1 "4-12C12]
achieve this inverse kinematics is employed. The A0 S124 C124 0 llS1 +/2SI2.1.
dynamics study is done using the Newton-Euler 0 0 1 -d2-d3
method. The significance of this study lies in the 0 0 0 -I
fact that it gives insight into the dynamic behavior
The ann matrix (T) can also be expressed in
of this robot.
terms of a hand coordinate system using n, s, a, p
notations where n is the normal vector, s is the
3.1. Kinematic analysis
sliding vector, a is the approach vector and p is
the position vector of the hand.
Robot kinematics deals with the analytical
study of motion of the robot ann with respect to
a fixed reference coordinate system as a function ny Sy ay py
of the time and without regard to the force or T= Sz az .
torque that causes motion. The kinematics can be 0 0
of two types: direct kinematics or inverse kine-
Comparing the T and A ° matrices we have:
matics. In the direct kinematics, the joint and link
parameters are given and the end effector posi- Px = 11C1+/2C12
tion is calculated. In the inverse kinematics the py = llS l +/2S12
joint angles are calculated for a given end effec-
tor positon and link parameters. In this work, the Pz= - d 2 - d 3 "
74 S.K. Padhy
C4 -$4 !]
0 2 = t a n - ' [ + ~/(1 - C22)//C2]
Ol = t a n - ' [ ( - M p . + N p y ) / ( M p y
d3 = - d2 - Pz,
+ Np.]
R 2=
[!o o] 1 0
where M = 12S 2, N = l I + 12C 2.
3.2. D y n a m i c analysis
C12--Sl2
C,2o
0 1
[C244
24]C124
motion with regard to the torques applied by the
actuators or by external forces applied to the
manipulator. There are two problems related to
the dynamics of a manipulator. The first one is R°4= SOo
direct dynamics where the torque is calculated
for given values of angular displacement, angular
velocity and acceleration. The second problem is
R[Cl Sly}
0= -S~ Cl
of indirect dynamics where the angular velocity,
0 0
angular displacements and acceleration are calcu-
lated for a given torque.
The dynamic study of a manipulator can be R2= -S 2 Co
done by several methods such as: Lagrange Euler
0 0
(L-E) method, Newton-Euler (N-E) method etc.
However, the L-E method is not commonly used
for real time control as it needs large amount of
computation time and space. The N-E method
involves a set of forward and backward recursive
R 2= [ C~2
-Sl2
0
Sl2
Cxz
0
0]
0
1
equations. The forward recursion propagates the R 3 = R2o .
kinematics information such as velocities and ac-
celerations at the center of mass of each link. The The vectors representing the distance from ( i -
backward recursive formulations propagates the 1)th co-ordinate system to ith co-ordinate system
forces and moments exerted on each link from are given as follows:
the end effector to the base of the robot. It has
observed that the computation time is linearly PI* = [/lCl, l l S l , 0] T
proportional to the number of joints and inde-
P2* = [11C12,/lS12, - d 2 ] T
pendent of the robot arm configuration. It is
feasible to implement this method in real-time p~ = [0, 0, - d 3 ] T
control of the robot. In this paper, the mathemat-
ical formulations of the equation of motion are P4* = [0, 0, 0] T.
derived using the N-E method.
Ro= C1
!]
The rotation matrices are as follows:
[c: s
The following initial conditions are assumed for
the present analysis:
to0 = tb0 = V0 = 0 .
0 Vo = (o, o, g)T.
On the dynamicsof SCARA robot 75
'2(0"1+0"2) +I1021S2+11(}'1C2,g] T
Rav3 = Ra(Zo(}'3 + R2I,;'2) + (Ra&3) × (RapS) R3a3 = (R~&3) × (R3e3) + (R3to3)
+ [2(so~) × (R~Zo0~)] × [(RSota3) × (R~oe,] + (R3I/3)
+ n~oJ z × [(R~oJs) × (R~p~)] = R3I,;'3
110"1C2.+1102S2-[-12{0"1-.k 0"2), g} T
{ l,0",(x + m,/2)
On the dynamics o f S C A R A robot 77
80.000
60.000
40.000 Lhnk 1
.~ 20.000
E
Z :J:ll .... ~ Link 2
0.000
D
Cr
©
~ -~0.000 V
-.40.000
-60.000