See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224606136

Accurate position control of PAM actuator in Lab VIEW environment

Conference Paper · October 2009

DOI: 10.1109/SISY.2009.5291145 · Source: IEEE Xplore

CITATIONS READS

18 254

4 authors, including:

József Sárosi Antal Véha

University of Szeged University of Szeged
88 PUBLICATIONS 572 CITATIONS 20 PUBLICATIONS 101 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by József Sárosi on 29 September 2014.

The user has requested enhancement of the downloaded file.

 Accurate Position Control of PAM Actuator in
LabVIEW Environment
J. Sárosi*, J. Gyeviki*, A. Véha* and P. Toman**
*
Faculty of Engineering, University of Szeged, Szeged, Hungary
** Alfa Busz Kft., Székesfehérvár, Hungary

[email protected], [email protected], [email protected], [email protected]

Abstract—As an important driver element, the pneumatic

artificial muscle (PAM) is widely used in industrial II. DESIGN OF A SLIDING MODE CONTROLLER
applications for many automation purposes thanks to their The pneumatic valve is the key element in the system.
variety of advantages. The design of a stable robust position There are two types of valves used in the pneumatic
controller for PAM is difficult since it is a very nonlinear positioning, servo-valves and ON/OFF valves. With
time-variant controlled plant because of the compressibility conventional ON/OFF valves accurate position control is
of air, air mass flow rate through the valve, etc. The main difficult to achieve because of the limitation of the valve
contribution of this paper is a robust position control response time. In the past few years there has been a wide
method based on sliding mode for a robot arm, driven by interest in the use of cheap high speed solenoid valves [7].
pneumatic muscle actuator. Finally, it presents experimental
The most of applications are on pulse with modulation
results.
(PWM). By the advent of DSPs with high computation
power, the precise and robust control of pneumatic
I. INTRODUCTION actuators has become possible [8].
This work is the first fundamental step of a wider Sliding mode control was introduced in the late 1970's
project, aimed at studying the humanoid robot. Muscles [9] as a control design approach for the control of robotic
only generate a force via contraction, i.e. a muscle can manipulators. Among experimental studies, a few
only ‘‘pull’’ and does not ‘‘push.’’ One muscle (agonist) succeeded in showing closed-loop system behaviour
contracts and simultaneously the other muscle relaxes which was predicted by the theory [10].
(antagonist, which increases in length), thus producing a Another solution is to employ the advanced nonlinear
force and motion on the mass. The same effect can be control strategies developed in recent years (soft
realized in a rotational sense by generating a rotation or computing) [11].
torque on the robotic joint through the contraction of the The design of a sliding mode controller consists of
agonist and relaxation of the antagonist muscle. three main steps. One is the design of the sliding surface,
Many researchers have investigated the precise position the second step is the design of the control which holds
control of pneumatic muscles during the past several the system trajectory on the sliding surface, and the third
years. Most of them dealt with the control of single or and key step is the chattering-free implementation. The
antagonistic pneumatic muscles. purpose of the switching control law is to force the
Due to the fact that the results obtained with a classical nonlinear plant’s state trajectory to this surface and keep
PI controller were not good, robust control techniques on it. When the plant state trajectory is „above” the
were considered. For pneumatic muscles, the application surface, a feedback path has one gain and a different gain
of different control techniques is found in the literature, if the trajectory drops „below” the surface.
but a good performance requires the use of robust or non- Consider a single-input, single-output second-order
linear control techniques. A variety of approaches, with nonlinear dynamic system (1).
varying success, have been attempted. PID control, neural
networks, and adaptive control, among others, have been y = f ( y , y ) + B ⋅ u (1)
utilized [1, 2, 3, 4, 5, 6]. While PID control is well known,
the results are particularly sensitive to errors in the
feedforward term. Adaptive and neural network control Where “y” is the output signal (position) of the
may be more robust, but suffer from slow convergence controlled plant, “u” is the control signal. If “yd” denotes
and long training sessions respectively. Thus, adaptive the desired value, then the error between the reference and
control is not well suited for the fast movements required system states may be defined as (2).
of an orthotic actuator. Analogously, neural control, with
its training workspace, does not handle unique or y e = yd – y (2)
unexpected situations well.
Therefore, a non-linear robust control technique, sliding
mode, was applied to design a position controller.

978-1-4244-5349-8/09/$26.00 ©2009 IEEE 301

A. Sliding surface design If s > 0 or s < 0 the control law should be selected in a
Classically, a scalar variable “s” is calculated as a way, which ensures (6).
linear combination of the error and its derivative (3).
− y − λ ⋅ ( f ( y , y ) + B ⋅ u ) < 0 or
(6)
s = y e + λ ⋅ y e (3) − y − λ ⋅ ( f ( y , y ) + B ⋅ u ) > 0

The simplest control law that might lead to sliding

Let s ( y e , ye ) = 0 define the sliding surface in the space
mode is the relay (7).
of the error state. The purpose of sliding mode control law
is to force the state trajectory of the error to approach the
sliding surface and then move along the sliding surface to u = δ ⋅ sign(s) (7)
the origin (Fig. 1).
C. Chattering free implementation
Chattering is the main problem of sliding mode control
s= y· e and chattering free implementation is the key step in
y design of a sliding mode controller. A quite general
e +λ · solution is that the relay (which changes its output value
ye =
0 suddenly) is replaced by a saturation function. There is a
boundary layer around the sliding surface where the
y ep y e0 ye control signal is changing continuously. If the system
trajectory is close to the sliding surface and the control
signal is small, than the system might stick before the
2 goal.
1
3 To avoid it a modified saturation function shown in
Table 1 is proposed. When the limitation of the position is
satisfied, all high-speed ON/OFF solenoid valves are ON
P to stop the overshoot. The control will be finished when
“‫׀‬yes‫ ”׀‬is smaller than “ye”. The modified saturation
function for the proportional valve is shown in Fig. 2.
Figure 1. Sliding motion in the state space

The process of sliding mode control can be divided into U

two phases, that is, the approaching phase with
s ( y e , ye ) ≠ 0 and the sliding phase with s ( ye , ye ) = 0 . F.FAST
Here 1 denotes the approaching phase, 2 and 3 denote
the sliding phase. If the system is in sliding mode the error
is decreasing exponentially, where “λ” is a time constant F.SLOW
type parameter. If “λ” is small than the system response is IN POSITION
slow but accurate. If it is big than the system response is s
fast but the system might chatter. B.SLOW

B. Selection of the control law B.FAST

In order to guarantee that the trajectory of the error
vector e will translate from approaching phase to sliding
phase, the control strategy must satisfy the sliding -s k2 -s k1 sk1 sk2
condition (4). Figure 2. Control function as modified saturation function

s ( y e , y e ) ⋅ s( ye , y e ) < 0 (4) III. THE SERVOPNEUMATIC POSITIONING SYSTEM

The experimental set-up is shown in Fig. 4, Fig. 5 and
This means that “ye” will always go toward the sliding Fig. 7 consists of a slider mechanism. One side of the
surface. A proper control should be selected to satisfy the muscle is fixed to a load cell, while the other side is
condition (4) in any time instant. Let us assume that the attached to the movable frame. The load cell (7923 type
desired value is constant and according to (2) and (3) from MOM) is a 4 bridge element of strain gauges. It is
follows (5). mounted inline to the PAM on the fixed surface. The load
cell measures the force exerted by the PAM. The linear
s = y e + λ ⋅ ye = − y − λ ⋅ y = − y − λ ⋅ ( f ( y, y ) + B ⋅ u ) (5) displacement of the actuator is measured using a
LINIMIK MSA 320 type linear incremental encoder with
0,01 mm resolution. Velocity and acceleration are
obtained by numerical derivation. During each test, slider
position, muscle force and applied gauge pressure are
recorded. Since PAMs are one-way acting, two are needed
to generate bidirectional motion: as one of them moves the

302
load, the other one will act as a brake to stop the load at its

Load Cell
Incremental Encoder
PC
desired position. To move the load in the opposite p2 x
C
p1

direction the muscles change function. The PAMs were p p

installed horizontally such that the only force present I I NI PCI-6251
M Series
during activation was the small friction force of the slider V2 V1
Multifunction
DAQ
mechanism. In the test-bed, two DMSP-20-200N-RM-RM V2a V2b V1a V1b
type fluidic muscle (from FESTO) can controlled by tree- SCB 68
Shielded I/O
way and two-way solenoid valves (MATRIX HX 751.102 Connector Block

C 324 3/2 NC and PX 861.9E4C2KK fast switching VB VA

types) and FESTO MPYE-5-M5-010-B type proportional PCM 8130 PCM 8130
valve. ps ps Driver Board Driver Board

We repeated experiments for several levels of pressures Driver

Board

in the range from 0 to 5 bar. To measure the air pressure,

Figure 5. Configuration of pneumatic positioning system with
two Motorola MPX5999D pressure sensors were plumbed ON/OFF valves
into the pneumatic circuit. A National Instruments data
acquisition card (NI 6251/M) reads the signal of force, The system pressure is set to be 5 bar, the sampling
pressure sensors and incremental encoder into the PC. time is 10 ms. In order to analyze the positioning methods
National Instruments LabVIEW will be used to monitor a real-time data acquisition program was designed. The
and collect the data imported through the DAQ card. It control program is based on Table 1.
will also dispatch the control profiles for the PAMs (Fig. 6
and Fig. 8). The testing system capable of recording the
TABLE I.
characteristic of PAMs (Fig. 3). CONTROL PROGRAM

Load cell
Fast Slow In Slow Fast
p=5.5 bar (constant)
2500
Forward Forward Position Backward Backward
VA 1 1 1 0 0
2000
VB 0 0 1 1 1
1500
V1a 1 1 0 1 1
V1
Force [N]

V1b 1 0 0 0 1
1000
V2a 1 1 0 1 1
V2
500
V2b 1 0 0 0 1

0
-10 0 10 20 30 40 50 60
Position [mm]

Figure 3. Force-contraction-pressure relationship for FESTO muscle

SCB 68
Valves
Load cell

Slider

Figure 6. Front Panel of the LabVIEW program for ON/OFF control

Incremental Encoder
Load Cell

Incremental Encoder
Muscles x PC
p2 p1
C
Figure 4. The photo of the experimental set-up
p p
I I NI PCI-6251
M Series
Multifunkc.
V DAQ

SCB 68
Proportional Valve
ps Shielded I/O
Connector Block

Figure 7. Configuration of pneumatic positioning system with

proportional valve

303
 Overshoot : 0.03 mm
Steady-state error : 0.00 mm
10.05

Position [mm]
10

9.95
1
VA
0
1
VB
0

Valves
1
Figure 8. Front Panel of the LabVIEW program for proportional V1a
control V2a
0
1
V1b
IV. EXPERIMENTAL RESULT V2b
0
0 1 2 3 4 5 6 7 8 9
The conventional, single stage solenoid operated Time [s]
ON/OFF valves are very bulky and their dynamic Figure 10. The time functions of the position and control signal
performances are low. With these valves fine motion (enlarged)
control is difficult to achieve because of the limitation of
the valve response time. With ON/OFF control the system Overshoot : 0.1 mm
Steady-state error : 0.01 mm
will never reach a steady state value.
15
The actual position will tend to oscillate around the
desired position. The second measurement is a positioning
Position [mm]

10
with high-speed ON/OFF solenoid valves. The time
functions of the position, and control signal is shown Fig. 5
9 and Fig. 10. The position error of the LabVIEW-based
relay type sliding mode control is within ±0.02 mm. The 0
next measurement is a positioning with a FESTO MPYE-
Control Signal [V]

5-M5-010-B type proportional valve. (Fig. 11 and Fig. 6

12).
5

Overshoot : 0.03 mm 4
Steady-state error : 0.00 mm
0 1 2 3 4 5 6
10 Time [s]

8 Figure 11. The time functions of the position and control signal with
Position [mm]

6
proportional valve

4 Overshoot : 0.1 mm
Steady-state error : 0.01 mm

2
15.1
0
Position [mm]

15.05
1
15
VA
14.95
0 14.9
1
14.85
VB
Control Signal [V]

7
Valves

0
1 6
V1a
V2a 5
0
1 4
V1b
V2b 0 1 2 3 4 5 6
0 Time [s]
0 1 2 3 4 5 6 7 8 9
Time [s]
Figure 12. The time functions of the position and control signal with
Figure 9. The time functions of the position and control signals with proportional valve (enlarged)
ON/OFF valves
This behavior is in absolute contrast to that of a
The experimental results show en excellent control pneumatic cylinder: a cylinder develops a force which
performance and that the sliding mode control is an depends only on the pressure and the piston surface area
effective methods to develop a practically available so that at a constant pressure, it will be constant regardless
human-friendly robot by using the PAM manipulator. of the displacement.

304
 V. CONCLUSIONS AND FUTURE WORKS [3] D. G. Caldwell, G. A. Medrano-Cerda, and M. Goodwin,
“Control of Pneumatic Muscle Actuator”, IEEE Control Systems
This work is the first fundamental step of a wider Magazine, pp. 40-48., February 1995
project aimed at studying the PAMs. With the help of this [4] G. A. Medrano-Cerda, C. Bowler, and D. G. Caldwell,
test-bed we can carried out several static and dynamic “Adaptive position control of antagonistic pneumatic muscle
investigations and control methods. Based on the actuators”, IEEE/RSJ International Conference on Intelligent
laboratory measurements we can conclude that the Robots and Systems, vol. 1, pp. 378-383., August 1995
pneumatic servo-systems can be used for precise robust [5] B. Tondu and P. Lopez, “Modeling and control of McKibben
artiffcial muscle robot actuators”, IEEE Control Systems
position control. The sliding mode control is a promising Magazine, vol. 20, no. 2, pp. 15-38, April 2000
tool for controlling such systems. The proposed modified
[6] F. Daerden, “Conception and Realization of Pleated Pneumatic
saturation function can eliminate the chattering, which is Artificial Muscles and their Use as Compliant Actuation
the main problem in case of sliding mode control. Elements”, PhD Dissertation, Vrije Universiteit Brussel, pp. 12-
Further works we have done with applying the input 19., 1999
shaping method. Once the system has reached the [7] M. C. Shih and M. A. Ma, “Position Control of a Pneumatic
setpoint, the residual oscillation will degrade positioning Rodless Cylinder Using Sliding Mode M-D-PWM Control the
High Speed Solenoid Valves” JSME International Journal Series
accuracy and may cause a delay in task completion. Input C, vol. 41, no. 2, pp. 236-241., 1998
Shaping is a feedforward control technique for reducing [8] P. Korondi and J. Gyeviki: “Robust and Precise Control for a
vibrations in computer controlled machines. The method Pneumatic Cylinder” EPE-PEMC, pp. 513 - 518., 2006
works by creating a command signal that cancels its own [9] V. Utkin, “Variable Structure Systems with Sliding Mode”
vibration. That is, vibration caused by the first part of the IEEE Transaction on Automatic Control, vol. AC-22, no. 2, pp.
command signal is canceled by vibration caused by the 212-222., 1977
second part of the command. Input shaping is a command [10] P. Korondi and H. Hashimoto, “Sliding Mode Design for
generation technique that is used to reduce command- Motion Control” Studies in Applied Electromagnetics and
induced vibration (as opposed to disturbance-induced Mechanics, vol. 16., p 12, 2000
vibration) [12]. Input shaping is implemented by [11] Gy. Mester, “Neuro-Fuzzy-Genetic Controller Design for Robot
convolving a sequence of impulses, called an input shaper. Manipulators.” Proceedings IECON’95, IEEE, Orlando, Florida,
USA, vol.1., pp. 87-92., 1995
[12] W Singhose, E. Biediger, Y. H. Chen and B. Mills, “Reference
REFERENCES command shaping using specified-negative-amplitude input
[1] T. Hesselroth, K. Sarkar, and K. S. Patrik van der Smagt, shapers for vibration reduction” ASME J. of Dynamic Systems,
“Neural network control of a pneumatic robot arm” IEEE Measurement and Controls, 2004
Transactions on Systems, Man and Cybernetics, vol. 24, no. 1, pp.
28-38., January 1994
[2] D. G. Caldwell, G. A. Medrano-Cerda, and M. Goodwin,
“Braided pneumatic actuator control of a multi-jointed
manipulator” Proceedings of the 1993 International Conference on
Systems, Man and Cybernetics, vol. 1, pp. 423-428., October 1993

305

View publication stats