Professional Documents
Culture Documents
476
CH2555-1/88/0000/0476$01.000 1988 IEEE
simplified model is sufficient but that both structural and joint
flexibility must be included.
(a) Schematic
wx;
\"I
0
IY
(b) Terminology Using the Assumed-Mode method [18], the solution of the
flexible motion can be represented as the linear combination of
Fig.1 A single degree flexible manipulator
admissible functions Qi(x) multiplied by time-dependent
generalized coordinates q,(t), where the admissible function
For small angular rotation 8, and deflection U ( u c e x l ), @i(X) should satisfy all the essential boundary conditions, that
we can assume is
Y = Z1@lql (10)
Then the kinetic energy T and potential energy V for such a
system with a payload Mp can be witten as In order to study the orthogonality of natural modes, the
external torque Tln is set to zero. The orthogonality
relationship of the second derivatives, which is true for a
system with only structural flexibility, will not hold here.
Instead, we have
(3)
JoLp@,Oldx + 1 ~ ~ , ( 0 ) ? ( 0 +
) M p O , ( ~ P l ( ~ ) =1. (constant) for I =I
(11)
The nonconservative work for input torque Tj,, is {O for I +I
W - T,,B (4) and
By using Hamilton's Principle [18], we have for
- k, [q(0)l2 I =j
, =
JoLEl O , @ dx
I q I,
'
- ki q ( 0 ) Q&O) for I ti
(1 2)
Substituting Eq. (IO) into Eqs. (2), (3) and using the
orthogonality relations Eqs. (1 1) and (12); the expressions for
kinetic and potential energy become
The rotatory inertia and shear deformation are more 1 n n
pronounced at the high frequencies and more influencial on the T = 2 l: zlq,ql [JoLPQ,@ldx + IH4(0)
@j(o) + M p W P , ( ~ ) ]
higher modes [ 171. Since the later results show that we can use
the first two modes in the dynamic modeling, we ignoring the
rotatory inertia and shear deformation effects. After
mathematical manipulation of Eq. 5 , we have the beam equation
411
and
n .
W = Tino = Tin ,Z, q ( 0 ) q, (15)
A14
0 IoL pO:dx + M,$W
0 0 0 0 TH
0 Kl 0 0 TK
The selected system outputs are the tip position, the angu!ar JoLEl&ldx+K,b:(0) 0
Q=
displacement and the strain, to be measured respectively by an 0
end point sensor, an angle sensor, and strain gauges mounted Lo 0 0 JoLEl&zdx + Klb$O)
immediately adjacent to the joint on the side of beam.The -0-
output vector, Z, can be expressed by state variables as
follows:
where
Results
In order to illustrate the significance of the dynamic models .m
developed in this paper, simulation of a single degree of 91
freedom manipulator, including both a flexible structure and a . lyx)
flexible joint, are shown below. A comparison among . loo0
different joint flexibilities is made in order to demonstrate the
importance of incorporating joint flexibility into the dynamic .o500
model. The examples in this paper are based on the decoupled n
dynamic model incorporating only the first two flexible 0 .loo0 .zoo0 .3om .4wo .sow .m
modes. To simplify the computation, the following parameters Ti" (Smcond)
0
0 ,1000 ,2000 ,3000 ,4000 ,5000 .SO00
TIM (Second)
Table 1 gives the system mode parameters hi based on Eq. Fig. 4. Second mode for three flexible arms
(30). We find from Table 1 that there is no zero mode (i.e. (case 1: X,; case 2,3: X,.)
rigid mode) for a flexible joint system (&# 0).
Table 1. System Hode Parameters 1, to3
1.200
.eo00
q3
,6000
.a
.zoo0
A package for computer aided design of control system,
MAControl [20], was used to simulate the QDen loop system 0
response for a unit step input in torque. The simulation results 0 . loo0 .2oo0 .3o00 .4oo0 .SO00 ,6000
are shown in Figures 3 through 14 . Figures 3 to 6 give the Ti.. (Second)
first four modes of the flexible arms for three different joint Fig. 5. Third mode for three flexible arms
stiffness coefficients. The first two mode shapes for these (case 1: X,: case 2, 3: X3.)
flexible arms are plotted in Fig. 7 and 8. Fig. 9 shows the
deflection of an entire flexible arm (Kc= 1) at t = 0.1,0.3,0.5
10-6
second, respectively. A comparison of the deflections for three 45 00 1 t
different flexible arms at t = 0.3 sec is given in Fig. 10. In 40.00
both cases, the deflection is computed based on the summation 35.00
over two modes U = Q1q1 + Q2q2. Figures 11 to 13 plot the 30.00
output variables; tip position, angular rotation, and strain vs. 2500
time. The tip deflection, the difference between the desired and q420 00
actual end-effector position due to flexilibities, is plotted in 15.00
Fig.14. The effect of the joint flexibility is shown in Fig.10
and 11. The deviation of curve 2 ( & = 1) from curve 1 (Kc = 10.00
0, rigid joint) in Fig.11 is approximately 50%. For curve 3 5.OOO
(& = 5 ) , the deviation from curve 1 increases to 80% at t = 0
0 .lo00 .2000 .)DO0 .4000 .SO00 ,6000
0.6 sec. Since a large amount of the input energy is stored in Ti- (Sec&>
the spring-like flexible joint, the deflection of the beam with Fig. 6. Fourth mode for three flexible arms
the flexible joint is much smaller than the one with the rigid (case 1: X,; case 2, 3: X4.)
joint, although it does have significantly more oscillatory
behavior.
419
*(XI
1 200
1 000
BOO0
6000
,4000
2000
0
0 2000 4000 6000 BO00 1 000
X (m)
Fig 7 First mode shape for three flexible arms
(case 1 Lo, case 2, 3 A, ) Fig. 11. Tip position for three flexible arms
(measured from initial coordinate system)
6000 I
I i I
4000
2000
4(X) 0
- 200
- 400
- 600
- 800
-1 00
I i !
0 2000 4000 6000 8000 1 000
X (m)
Fig 8 Second mode shape for three flexible arms
(case 1 A,: case 2. 3 A, ) Fig. 12. Angle e for three flexible arms
0 100
- 010
- 020
- 030
-MO ,
I
I I I -1 575 0 .loo0 ,2000 ,3000 .4000 .Moo 6000
0 2000 4000 6000 BOO0 1 000 Tb-M (Srcnd)
i I ,0200
0
.0100
- ws 0
-.OlO
- 010 - ,020
-.015
.__ 3 Kc-5.
-.mo
-.MO
- ---
mn -.OS0
I I I I I 1-1 .%9
48 1