Professional Documents
Culture Documents
1868
( 3 ) The outputs of the fuzzy links of PUMA 5 6 0 . Table 4.1 shows
controller are joint torque 1 and the geometric and inertia parameters
joint torque 2. We choose their of the robot [ S I . The membership
membership functions to be the same functions for PUMA 5 6 0 are presented
and depict it in Figure 3 . 3 . in Figure 4 . 2 through Figure 4 . 4 . By
using the same three steps as we did
The rules in . (2) are for a two-link manipulator, we
implemented with' the help ofsthe design the fuzzy controller for PUMA
personal computer and FIDE software. 560.
The fuzzy controller fuzzifies the
input quantities through algorithms
that would operate on the input data 5. SOFTWARE IMPLEMENTATION AND
as specified by the membership SIMULATION RESTJILTS
functions described in (1). The IF-
THEN decision rules can be Fuzzy Inference Development
implemented using Table 3.1. And Environment (FIDE) software by
finally the output is defuzzified Aptronix, Inc. :is utilized for
based on the membership functions implementing the fuzzy if-then rules
described in ( 3 ) . [ 6 1 . The two-link manipulator is
The whole procedure is a standard simulated by code written in c
one for fuzzy controller designs language [ 7 1 . The simulation of PUMA
5 6 0 is derived as an extension to
[1,21.
the simulation of the two-link
manipulator. Space constraints force
4. FUZZY CONTROLLER DESIGN FOR the exclusion of the code in this
ROBOT PUMA 560 paper, but the reader can contact
the authors for more information on
PUMA 5 6 0 is a popular this subject.
industrial robot. Its
characteristics are as follows: The previous sections describe
the membership functions for the
* It is a medium-power robot; various input arid output variables
* It is programmable either for of the fuzzy coritroller. Nothing is
point-to-point or for continuous and perfect in this world, and fuzzy
is computer controlled; control tries to explore this
* It comes with a teach pendant and approximate, inexact nature of the
is powered by DC electric motors; real world. A set of rules are
* It is widely used in manufactur- written for the robotic systems and
ing process, for handling small simulated in FIZ)E to give outputs.
parts ; Application of robots in industry
* It is actively used in and other fields: depends on
educational institutions for efficiency, reliability and the
research purposes. capabilities of the control system,
which has to ensure successful
We assume the first three links of application of robots in various
PUMA 560 have the same membership tasks. The control system of robots
functions while the last three links can be realized in different ways,
are locked and therefore become the with varying degrees of complexity
load of the first three links. depending on the tasks imposed upon
Figure 4.1 depicts the typical a specific robot. Fuzzy logic
coordinate frame assigned to the control can be considered as a nexus
between the conventional precise
1869
mathematical control and the human- the conventional controller in the
like decision making schemes. At sense that the position trajectories
this time there is no systematic of the robot joints with the fuzzy
procedure for the design of a fuzzy controller is much closer to the
controller and heuristic rules are desired trajectories.
used for the control scheme. (2) In Figure 5 . 3 through Figure
Linguistic rules are the heart of 5 . 4 and Figure 5.12 through Figure
the fuzzy implementation. By a 5.14, the desired trajectories
linguistic variable we mean a remain the same. The trajectories of
variable whose values are words or the robots with the fuzzy controller
sentences in a natural or artificial when loaded still go convergent to
language. For example, Age is a the desired trajectories, while the
linguistic variable if its values conventional control plots show a
are linguistic rather than considerable discrepancy with the
numerical, i.e. young, not young, desired plots.
very young, quite young, old, not (3) Figure 5 . 5 through Figure 5 . 6
very old, and not very young etc. and Figure 5 . 1 5 through Figure 5.17
rather than 2 0 , 2 1 , 2 2 , 2 3 . . . depict the situation when the
desired trajectories are sinusoidal
Using the fuzzy linguistic functions. Evidently, the
rules, various conditions are conventional controller can hardly
simulated to test the validity of function then. The fuzzy controller
the fuzzy controller. The behavior is still working very well to bring
of the two-link manipulator as well the actual trajectories of the
as that of PUMA 5 6 0 are demonstrated robots to the desired trajectories.
for either the fuzzy controller (4) In robotic grinding, deburring
applied or the conventional (PD and or assembly, smooth transition from
PID) controller applied. The results free to constrained motion is of
obtained for the two-link special interest. Figure 5 . 7 through
manipulator are presented in Figure Figure 5 . 8 and Figure 5.18 through
5.1 through Figure 5.8 in the form Figure 5.20 present the results
of plots of joint positions versus obtained in these cases. Again, the
time. Figure 5 . 9 through Figure 5 . 2 0 fuzzy plots are convergent to the
present the simulation results for desired plots. The conventional
PUMA 5 6 0 . Basically four different plots now show a total divergence.
conditions are experimented
respectively: NOTE: Due to limitation of space,
the simulation results cannot be
. Robot Unloaded
Robot Loaded
presented in this paper.
1870
links locked. The performance of
these robots with a fuzzy controller
applied is found to be better than
with a conventional controller
applied. Moreover, this new approach
is easier to implement. Simulation
results for various conclitions for
the robots find that the fuzzy
controller provides a robust
control.
REFERENCES
Table 3.1 Tbe Rule Base for llhe fuzzy kmowlcdge base matrd
1. D. Driankov, H.Hellendoorn and M.
i
Reinfrank, An Introduction to Fuzzy Control,
NegativeSmall PositiveSmall
Springer-Verlag, 1993
2. T .J . Ross, Fuzzy Logic with Engineering
NegativeLarge
Application, McGraw Hill, 1995
3. J.J. Craig, Introduction to Robotics - NegativeSmall Zero NegativeSmaU NegahMdium
Machanics and Control, Addision-Wesley, 1989
4. H. Asada and J.E. Slotine, Robot Analysis Zero FmithSmall NegativeSmaIl
and Control, John-Wiley and Sons, 1987
5. T.J. Tarn, A.K. Bejecy and X. Yun, ZWJ
“Dynamic Equations for PUMA 560 Robot Arm”,
Dept. Of Systems, Science and Mathematics, RxitkSmall
1871
0.7764
1872