Proceedings of the 2010 IEEE

International Conference on Information and Automation

June 20 - 23, Harbin, China

High Performance Robotic Gripper Based on Choice of

Feedback Variables

Aziza. M. Zaki and Ayman. M. Soliman Osama. A. Mahgoub and Abdellatif.M. El-Shafei
Power Electronics and Energy Conversion Department Electric Power and Machines Department
Electronics Research Institute Faculty of Engineering, Cairo University
Cairo, Egypt Cairo, Egypt
[email protected] [email protected]
[email protected] [email protected]

Abstract - The human operators may be subjected to of non-contact sensing is vision sensing where video camera
hazardous environments or monotony routine work in pick and is processed to give the robot the object location information.
place type operation. One has to rely on robotic or autonomous However, it is costly and gives no data concerning forces [3].
systems that have robust and flexible mechanical structure to Tactile sensing, on the other hand, has the capability to do
adapt to the needs of environment in order to remove these
proximity sensing as well as force sensing, it is less
hazardous. The gripper is one of a critical component of an
industrial robot which is often useful in industrial environments expensive, faster and needs less complex equipment [4-6].
for object grasping during handling process. The applied force The basic principle of the Slip-Sensitive Reaction used in this
by the gripper on the grasped object should be well controlled to work is that, the gripper should able to automatically react to
avoid the risk of the object slipping as well as any possible an object slipping from its grasp with the application of
damage to the object. This goal can be difficult to achieve greater force. A lot of researches have been focused on
because of the nonlinearities such as mechanical system and fingertip sensors developments to detect slippage and applied
nature of used sensors and friction forces. Researchers in this force [7-10], which require complicated drive circuit and
field handle the gripper from several points of view such as the suffers from difficult data processing and calibration.
construction effect that appear in two fingers or multi fingers
Polyvinylidene fluoride (PVDF) piezoelectric sensors are
configuration and sensors investigation to increase the robot
awareness and the control strategies. Robotic gripper and the presented in [4-6] to detect contact normal force as well as
criteria of configuration are investigated in this paper. The slip. Also, an array 8x8 matrix photo resistor is introduced in
importance of ANFIS modeling of the system to be closed to real [11] to detect slippage. A slip sensor that is based on the
system is presented. The performance of this system is enhanced operation of optical encoder is used to monitor the slip rate as
by proper choice of the feedback variables in the proposed a result of insufficient force is presented in [12]. However, it
control strategy. The proposed robotic gripper and control is costly and gives some object constrains. Several researchers
strategy is verified experimentally. handle finger adaptation using more than one link in one
finger to verify stable grasping [13-15]. This results in
Index Terms - Intelligent Robotic Gripper, ANFIS Modeling, complicated mechanical system leading to difficulty in
Fuzzy Control and Grasping Algorithm
control and slow response. In [16], variable reluctance
technology as an actuator is used in two-finger gripper that is
I. INTRODUCTION
inherently nonlinear. Stepper actuator is used in [14] where
Robotic system has become mandatory in our life, the finger position depends on the step value. Fuzzy
especially if humans are exposed to hazardous environments. controllers have been very successful in solving the grasping
Robotic gripper is crucial in these areas. Generally, the main problem, as they do not need the mathematical model of the
goal of robotic gripper during object grasping and lifting system [17]. All previous work used a combination of applied
process is applying sufficient force to avoid the risk of the force, applied force rate, object slip rate, and its derivative as
object slipping as well as any possible damage to the object. feedback variables for their control strategies. In this paper,
Until now, this goal is difficult to be achieved because of the we present a new design to and implementation of robotic
non-linearity in the system such as mechanical system gripper with electric actuation using brushless dc servo
configuration, nature of used sensors and friction forces, and motor. Standard sensors adaptation in this work leads to
it can be posed as an optimization problem [1, 2]. Sensory maintain the simplicity of the mechanical design and gripper
systems are very important in this field. Two types of sensing operation keeping a low-cost layout. System modeling had
are most actively being investigated to increase robot been introduced using ANFIS approach. Two new controller
awareness: contact and non-contact sensing. The main focus schemes based on human skill experience with different feed-

978-1-4244-5704-5/10/$26.00 ©2010 IEEE 54

back variables are presented and discussed. The results III. ROBOTIC GRIPPER MODELING USING ANFIS
proved that the performance is enhanced when the feedback
variables are well selected. In section II, a gripper To build the proposed controller, we need to know about
configuration is presented. Section III presents the gripper the proposed system characteristics and modeling for
modeling using ANFIS. The object modeling is presented in simulation and experimental work purpose. Hence,
section IV. Proposed scheme controller for the gripper system input/output variables of the system are measured and
is developed in section V. Experimental results are illustrated processed according to the proposed configuration and
and discussed in section VI. Finally, section VII gives the gripper prototype as shown in Figure 2. The input variable to
conclusions of the work. the system is the speed control command for servo motor
driver and is expressed as reference voltage Vref . The applied
II. ROBOTIC GRIPPER CONFIGURATION
force on the object is the output variable Fapp . Hence, we
A proper gripper design can simplify the overall robot need to model the variation of the applied forces Fapp by the
system assembly, increase the overall system reliability, as
gripper finger with time at different reference voltage control
well as decrease the cost of implementing the system [18-20].
Hence, the configuration of the gripping system is very values Vref . Adaptive Neurofuzzy Inference Systems ANFIS
important for the successful operation. The developed gripper are realized by an appropriate combination of neural and
device was configured with a two finger parallel design for its fuzzy systems and provide a valuable modeling approach of
wide applications in spite of its precise need control. One complex systems [26]. Hence, input/output variables of the
finger is fixed and the other is movable to ease the control system are measured and processed. The ANFIS structure is
and minimize the cost as shown in figure 1. The fingers are applied on our proposed robotic gripper based on the
flat and rectangular in shape. The housing of gripper and measured data and simulated using MATLAB software
fingers were made of aluminum sheet for light weight package as shown in Figure 3. 161 training data sets, 46
consideration with proper thickness to ease the machining checking data sets, and 46 testing data sets are used. The
and holes puncture through edges for simple assembly and surface rules viewer for the developed FIS model using
ease in maintenance. The movable finger is derived on a lead ANFIS is shown in Figure 4. Simulation results of the gripper
screw, guided by a linear bearing system with its advantages ANFIS modeling is shown in Figure5.
of self-locking capability, low cost and ease of manufacture
[21]. Commercial force sensor resistor FSR, Flexiforce A201
standard with range of 0-1 lb (4.4 N) [22] is used to measure
the applied force. Phidget vibrator sensor [23] is adapted as
slip sensor to give information about object slip rate in m/sec.
Permanent magnet brushless dc motor is used as an actuator
to drive the movable finger due to its high power density, ease
of control, high efficiency, low maintenance and low rotor
inertia [24]. Hence, BLDC servo motor, internal rotor motor,
is used for the movable finger [25]. Long fingers lengths
require high developed torque and short lengths make
restriction in dimensions, hence fingers are selected to be 15
cm Length. Also, we consider a proper selection of contact
rubber material area between the fingers and object to
decrease the pressure on the object, increase the friction, and
to avoid deformation from centric concentrated force.
Fig. 2 Gripper prototype

Fig.3 Robotic gripper using ANFIS

Fig. 1 Robotic gripper configuration

55
 Fig. 4 Surface rules viewer for the developed FIS model using ANFIS

Fig. 6 Object simulation results when

Mass=100 gm and μ =0.5

V. GRIPPER SYSTEM CONTROLLER

By observing and investigating how the human being

handles unknown object, our proposed controllers are
developed. First, human touches the object to examine its
temperature and stiffness. Then, he tries to lift it by applying
small force to move it or lift it for acquiring some information
about its weight. Finally, he estimates the force needed to lift
this object and takes the decision if he can lift it or not. Based
on these observations, two schemes of control are developed
but with different feedback variables.
Fig.5 Gripper simulation results using ANFIS modeling

A. First Scheme Proposed System Controller

IV. OBJECT MODELING
During object grasping and lift process, it is not
The occurrence of slip for solid objects during grasping guaranteed that the two fingers will be in contact to the object
and lifting mainly depends on its mass and its coefficient of at the same time. Hence, a pushing force will be applied by
friction and finally the applied forces. And if the applied one finger (movable finger) before complete contact. This
force is not enough , acceleration is generated which leads to force will move the object. Normally, this pushing force is
increased rate of slip and object dropping after certain time less than the force needed to lift the object. Figure 7 shows
depending on the applied force, object mass and the the block diagram of the first proposed scheme controller.
coefficient of friction. Equation 1 determines the object Two integrated fuzzy controllers were built in this scheme as
acceleration as a function of the normal applied forces by the following:-
gripper fingers and the coefficient of friction. Object
simulation result is shown in Figure 6, which indicates that 1. The first fuzzy controller is a reference voltage
the slippage is stopped after a period of time depends on the controller with two input variables, slip-rate and
rate of force increase. its derivative.
m × a = m × g − 2 × μ × Fapp (1) 2. The second fuzzy controller is a gain controller
Where m is the object mass μ is the coefficient of friction, g for the output of the first controller with one
is the earth gravity, and a is the object acceleration. input variable, pushing force.

The function of the second controller is to decrease or

increase the reference voltage command. The output of this
controller is based on the pushing force applied on the object

56
before grasping and lifting process. Figures 8, 9 show the
surface viewer for the two controllers in this scheme.
Simulation results show the response of this scheme as shown
in Figure 10.

Fig. 7 Block diagram of the first scheme controller

Fig. 10 System response for the first scheme controller

mass = 300 gm and F push = 150 gm-force

B. Second Scheme Proposed System Controller

Three integrated fuzzy controllers were built in this

scheme as shown in Figure 11:-
Fig. 8 Surface viewer of the reference voltage controller
1. Guess starter reference voltage controller
2. Increased percent controller for starter reference
voltage command
3. Enhancement Controller for the Starter
Reference Voltage Command
The first controller is to guess the acceleration of the object
resulting from few applied force and gives the suitable value
of reference voltage command, the second controller is to
sense the pushing force to the object before the grasping
process and its output is multiplied by the first controller
output, the third controller is to enhance the behaviors of the
first and second controllers based on the object acceleration
and the applied force feed-back. The controllers receive the
object acceleration, object acceleration rate, pushing force and
Fig. 9 Surface viewer of the gain controller
the applied force as feedback variables and adjust the finger
motion. The response of this scheme is shown in Figure 12
which indicates a faster response and lower slippage than the
first scheme controller.

57
 system response. The comments inside figure explain the
scale conversion values obtained from sensors calibration.
Figure 13 shows the system response at lifting process when
object was 1000gm. Figure 13(a,b) show a good performance
in spite of the start reference controller based on pushing
force is not considered. Considering the start reference
controller based on pushing force is shown in Figure 13(b)
which we can minimize the time of the grasping process at
lifting, moreover the lower slippage reduction can be
achieved.

Ch4: 2.5V= 0.0 mm/s Slip rate= 0.0m m/s

0.8V=3.0 m/s Slippage is stopped

Ch3: 3.2V=1050 gm-force

Fig. 11 Block diagram of the second scheme controller 1.2V=300 gm-force

Ch1: 0.0V= 0.0V

500 mV=500 mV

Ch4: Slip-rate (mm/s) – Ch3: Fapp (gm-force) – Ch1: Vref (V)

(a) Pushing force is not considered

Fig. 13 System response when mass=1000gm

Ch4: 2.5V= 0.0 mm/s Slip rate= 0.0 mm/s

1.9V=1.0 m/s Slippage is stopped

Ch3: 3.2V=1050 gm-force

Fig. 12 System response for the second scheme controller 1.2V=300 gm-force
mass = 300 gm and F push = 150 gm-force

Ch1: 0.0V= 0.0V

VI. EXPERIMENTAL RESULTS 500 mV=500 mV

Experimental work were established to verify our goal

including designing, manufacturing, testing and fining the
proposed mechanical system, sensors calibration and building
its necessary drive circuits, recognizing on the actuator
Ch4: Slip-rate (mm/s) – Ch3: Fapp (gm-force) – Ch1: Vref (V)
characteristics and performance, and finally, using the micro-
computer to interface with the proposed gripper and writing (b) Pushing force is considered
the developed algorithm according to the second scheme Fig. 13 System response when mass=1000gm
controller. Experimental results were obtained for different
values of object mass with and without considering the
pushing force in control algorithm. Figure 13 shows the

58
 [14] C.M.Seguna and M.ASaliba,"The mechanical and control system design of
a dexterous roboticgripper", The 8th IEEE International Conference on
VII. CONCLUSIONS Electronics, Circuits and Systems, ICECS 2001, Vol. 3, pp. 1195 - 1201.
[15] V. Dubey, and R. Crowder, "Grasping and Control Issues in Adaptive End
In this paper, intelligence in designing and implementing Effectors", in ASME Design Engineering Technical Conferences, Salt Lake
City, Utah, September 28 - October 2, 2004.
a robotic gripper appears at handling unknown objects with [16] K.K.-C. Chan, and N.C. Cheung," Grasping of delicate objects by a novel
different masses, dimensions, and coefficients of frictions. All two-finger variable reluctance gripper", Thirty-Sixth IEEE Industry
items in gripper construction had been modeled. In the mean Applications Conference, IAS Annual Meeting Conference, Vol. 3, 30 Sep-4
Oct. 2001, pp.1969 – 1974.
time, two schemes of control with different feedback variables [17] J. A. Dominguez-Lopez, and D. N. Vila-Rosado," Hierarchical fuzzy
are tested. Owing to the proper choice of feedback variables, control to ensure stable grasping", Seventh Mexican International
system response and performance are enhanced as discussed Conference on Computer Science (ENC'06), Sept. 2006, pp. 37 – 43.
[18] R. G. Brown and R. C. Brost, “A 3-d modular gripper design Tool”, IEEE
in the second scheme controller compared to the first scheme
International Conference on Robotics and Automation Proceedings, 1997,
controller. Moreover, a new algorithm similar to human pp. 2332-2339.
behavior for the grasping process was presented. Simple rule [19] G. J. Monkman, S. Hesse, R. Steinmann,and H. Schunk, “Robot Grippers”,
base was used. Experimental results were presented and Wiley-VCH, 2007.
[20] J. W. Soto Martell, and G. Gini, “Robotic hands: design review and
discussed. The results show the fast time response in stopping proposal of new design process”, Proceedings of World Academy of Science,
the slippage. Engineering and Technology, Vol. 20, April 2007, pp. 85-90.
[21] J. E. Shigley, C. R. Mischke, and R. Gordon, “Mechanical Engineering
Design”, (7th ed.), McGraw Hill, 2003.
[22] Flexiforce sensor, www.tekscan.com
REFERENCES [23] Vibrator sensor, www.phidget-sensor.com
[24] Bhim Singh, A.H.N.Reddy, and S.S.Murthy, "Hybrid Fuzzy Logic
Proportional Plus Conventional Integral-Derivative Controller for
Permanent Magnet Brushless DC Motor", Proceedings of IEEE International
[1] E. Ottaviano, M. Toti, and M. Ceccarelli, "Grasp force control in two-finger Conference on Industrial Technology ICIT 2000, vol.1, 19-22 Jan. 2000,
grippers with pneumatic actuation", Proceedings of the 2000 IEEE pp. 185- 191.
International Conference on Robotics and Automation, San Francisco, USA, [25] Minimotor SA CH-6980 Croglio, Switzerland, Faulhaber dc Motors,
Vol.2, April 2000, pp. 1976 – 1981. www.minimotor.ch.
[2] A. Bicchi and V. Kumar, "Robotic grasping and contact: A review", [26] A. Rezaeeian, A. Y. Koma, B. Shasti, and A. Doosthoseini, “ANFIS
Proceedings of the 2000 IEEE International Conference on Robotics and modeling and feed forward control of shape memory alloy actuators”,
Automation, San Francisco, CA, 2000, pp. 348–353. International Journal of Mathematical Models And Methods in Applied
[3] R. D. Lorenz, K. M. Meyer, and D. M. Van de Riet, "A novel, compliant, Sciences, Issue 2, Volume 2, 2008, pp. 228-235.
four degree-of-freedom, robotic fingertip sensor", IEEE Trans. on Industry
Applications, vol. 26, no. 4, July/August 1990, pp. 613-619.
[4] B. Choi, H. Choi, and S. Kang," Development of tactile sensor for detecting
contact force and slip", IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS 2005), 2-6 Aug. 2005 , pp. 2638- 2643.
[5] G. Cannata and M. Maggiali, " An embedded tactile and force sensor for
robotic manipulation and grasping", 5th IEEE-RAS International
Conference on Humanoid Robots, Dec. 2005, pp. 80- 85.
[6] M. F. Barsky, D. K. Lindner, and R. 0. Claus, "Robot gripper control system
using PVDF piezoelectric sensors", IEEE Trans. on Ultrasonic,
Ferroelectrics, and Frequency Control, vol. 36. no. 1, Jan. 1989, pp.129-
134.
[7] R. Masuda et al., "Slip sensor of Industrial Robot and Its
Application."Electrical Engineering in Japan, vol.96, no.5, 1976, pp.129-
136.
[8] P. Dario and D. De Rossi, "Tactile sensors and the gripping challenge".IEEE
Spectrum, no 22, vol. 5, August 1985, pp.46-52.
[9] R.D. Howe and M.R. Cutkosky, "Sensing skin acceleration for texture and
slip perception" Proc. of the IEEE International Conference on Robotics and
Automation, Scottsdale, Arizona, May 14- 19,1989, pp.145-150.
[10] W. Friedrich, P. Lim, and H. Nicholls, " Sensory gripping system for
variable products", Proceedings of the 2000 IEEE International Conference
on Robotics and Automation, San Francisco, California, USA, April 2000,
pp. 1982-1987.
[11]Ren C. Luo, Kuo L. Su and S. H. Henry phang, "An implementation of
gripper control using the new slipping detector by multisensor fusion
method", Industrial Electronics Society, IECON 2000, 26th Annual
Confjerence of the IEEE, vol.2, pp. 888 - 893.
[12]M-J. E. Salami, N. Mir-Nassiri and S. N. Sidek," Design of intelligent
multifinger gripper for a robotic arm using a DSP-based fuzzy controller",
Proceedings of TENCON 2000, Kuala Lumpur, Malaysia, vol. 3, 24-27
Sept. 2000, pp. 348-353.
[13]A. M. DOLLAR. and R. D. HOWE, "Towards grasping in unstructured
environments: grasper compliance and configuration optimization"
Advanced Robotics, VSP and Robotics Society of Japan, Vol. 19, No. 5, pp.
523–543.

59