Proceedings of the ASME 2014 International Mechanical Engineering Congress and Exposition
IMECE2014
November 14-20, 2014, Montreal, Quebec, Canada
FIVE BAR PLANAR MANIPULATOR SIMULATION AND ANALYSIS BY BOND
GRAPH
Shengqi Jian
Faculty of Engineering and Applied Science,
Memorial University of Newfoundland
St. John’s, Newfoundland and Labrador, Canada
Luc Rolland
Faculty of Engineering and Applied Science,
Memorial University of Newfoundland
St. John’s, Newfoundland and Labrador, Canada
ABSTRACT
This work focuses on the bond graph modelling method
and its application on multi-body system, especially on the
five-bar parallel robot. Five-bar parallel robot is comprised of
four arms, two revolute actuators and five revolute joints. This
paper adopts five-bar parallel robot in symmetric configuration
as simulation object. As it will be used as a pickup and placing
machine, its workspace is fixed on Cartesian coordinate. The
relationship between the two rotating angles and end effector's
desire position is built by inverse kinematics. Bond graph is
used to describe moment, torque, velocity, angle relationships.
In this project, the dynamic performances between arms,
motors at robot basement and end effector will be researched.
In this paper, an investigation about how to use bond graph to
model DC (direct current) servo motor and an integrated
motion control system is carried out. During a typical end
effector point-point displacement, the torque change between
arms is plotted. Finally, 3-D animation experiment is
conducted. Experiment results show that bond graph can
simulate robot dynamics performance without having to make a
large number of equations. It is able to simulate and solve five-
bar kinematics problem in the process.
1. INTRODUCTION
The need for high speed and high accuracy has justified.
The application of parallel robots in areas such as aerospace,
precise machining, assembly line tends to be controllable close
chain mechanism. These mechanisms attract more attention
from researchers.
Parallel robot was introduced and researched decades ago.
They were firstly applied as air plane simulators and
1 Copyright © 2014 by ASME
Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
IMECE2014-37602
Cheng Yin
Faculty of Engineering and Applied Science,
Memorial University of Newfoundland
St. John’s, Newfoundland and Labrador, Canada
Lesley James
Faculty of Engineering and Applied Science,
Memorial University of Newfoundland
St. John’s, Newfoundland and Labrador, Canada
entertainment devices in cinemas. James. E. Gwinnett firstly
designed and patented a spherical parallel mechanism. In 1954,
Dr. Eric Gough designed and built the first octahedral hexapod
mechanism. It was used for tire testing and this mechanism was
duplicated until present. Klaus Cappel later revised parallel
mechanism as flight motion simulator [1]. Parallel mechanisms
compensate serial mechanisms' insufficient. It compromised a
more rigid structure with many advantages that serial robots do
not have. Parallel robots own high stiffness, relatively large
payload capacity and low inertia. However, it also has some
drawbacks, such as limited workspace and nonlinear dynamics
leading to control hardships.
Five bar planar manipulator is a relatively simple
mechanism and its kinematics is explicit. However, its
characteristics are high speed, high accuracy, low inertia and
high stiffness. It has 2 degree-of-freedom (DOF). For these
reasons, it draws a lot of researchers’ attention. Some
prototypes and commercial products were made, such as
'double SCARA' RP-AH series offered by Mitsbishi Electrics,
and DexTAR, five bar planar manipulator designed by ETS.
With price decreasing, it has the potential to be widely used in
production line for picking and placing tasks and elements
assembly tasks.
In reference [2], It studies five bar manipulator bond graph
model. While it is composed of four revolute joints, one
prismatic joint and two actuators, main motor is driving crank
and another motor is driving linear power screw. In reference
[3], Dr. Karnopp and Dr. Margolis studies five bar mechanism
bond graph model, however, it only researched building
mechanism model without control or kinematics. This paper
simulation object is an implementary parallel robot system,
 which is engineering related. Its workspace is Cartesian labeled as L3 and L4. Two active joints are connected with two
coordinate. fixed actuators, labeled as A1 and A2. Three passive joints are
Nowadays, mechanism dynamic analysis mainly adopts the labeled as B1, B2 and P. It contains two DOFs on XY plane.
closed vector polygon method, the complex number method, or End effector is settled at P joint.
matrix method to build mechanism dynamic description. Then Inverse kinematics is used to derive the actuator angles
use methods of vector mechanics or analytical mechanics, such from end effector position. Forward kinematics is used to
as Lagrange–d'Alembert principle, Hamilton principle, calculate end effector position from actuator angles.
Wittenburg method or Kane method, to solve the dynamic
problems. Adopting these methods, the whole system dynamic
functions must be built. As for complex rigid coupling
mechanism, especially multi-loop, multi-bar mechanisms,
whole system dynamic functions are difficult to be built.
Bond graph is a method to describe system power
exchange, transmission, storage and dissipation. By introducing
real parameters and variables, it can objectively show the
relationship between all variables and get explicit state
equations. This method was firstly raised by Dr. Paynter [4] in
1960. Then Dr. Karnopp and Dr. Rosenberg used bond graph
simulating different mechanisms, electrical devices and
hydraulic mechanism. As systems have power transmission,
they are qualified to have bond graph model [5].

Fig. 2: Sketch of five-bar robot (in - + configuration)

As figure 3 shown, if one point is defined as the desired

point on the XY plane, usually five-bar parallel robot can have
four configurations to reach that point, as + +,+ -, - +,- - [9]. In
this work, the - +configuration is chosen.
Fig. 1: DexTAR robot, ETS, Montreal [6]

Figure 1 is a typical five bar planar robot, DexTAR (photo

courtesy of Dr. Bonev). It is designed by École de technologie
supérieure (ÉTS), Montreal [6].
Its main task is to proceed with object palletization. So
how to derive equation between end effector desired point and
desired motor angle, how to design controller to send motors to
set points, and how to build relationship between arms and
joints by bond graph, these questions are very critical to this
project and all waiting to be solved. (a) - + configuration (b) - - configuration

2. INVERSE KINEMATICS
As figure 2 showed, the planar five-bar parallel mechanism
compromises two actuators, four arms and five revolute joints.
The workspace, parallel singularities and serial singularities are
all influenced by four arm lengths ratio and distance offset
between two actuators [7]. In order to at best avoid singularity
points, symmetric configuration is chosen with two equal long
forearms and two equal short arms. Symmetric structure is
chosen in proposed design approach also due to the benefits of
workspace symmetry and simpler kinematics equations [8]. (c) + + configuration (d) + - configuration
Proximal arms are labeled as L1 and L2. Distal arms are Fig. 3: Five-bar four configurations

2 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 In this section, hereafter kinematic derivations obey ground Using inverse tangent value to calculate five-bar
reference frame shown in figure 2. The end effector's position is inverse kinematics, however, this approach is not accurate
defined as enough. During the calculation, some points on XY plane can
P x , y (1) lead to equaling to infinity. So here is chosen
instead of , it will not turn to infinity resulting program
In the X-Y base reference coordinate, and
collapse.
positions are shown in (5)(6). and positions are shown
Parallel robot’s inverse kinematics problem is easier to be
as following (7)(8). , is the length of proximal arms and solved than forward kinematics problem. Multiple solutions
they all equal to . , is the length of distal and they all might occur. Because reaching the same point on X-Y plane
equal to . is motor one’s angle. is motor two's angle. leads to different configurations. In this case, four
is half distance between two motors and it is set as . configurations are caused by multiple solutions.
r |A B | |A B | L L (2)
3. BOND GRAPH MODELING
r |B P| |B P| L L (3) Planar mechanism is simulated using bond graph by other
| | researchers. In Karnopp and Margolis paper, they described
r (4)
planar mechanism referring to bond graph [3]. In Ashraf’s
A r ,0 (5) paper, although he focused on simulate joint effects on
mechanism, there are still many methods to solve bond graph
A r ,0 (6) problems in it [10]. A bond graph model is particularly useful to
describe systems in which a variety of elements in different
B L cos θ , L sin θ (7)
energy domains interact [13]. Robot system is constituted by
mechanical, electrical, magnetic components.
B L cos θ , L sin θ (8)
Model of Actuators
r r 250 (9)
As figure 2 shows, it is the sketch of a DC motor. It
r 25 (10) comprises inductance , resistance , motor constant n,
rotation inertia J, bearing friction b and gear header constant m.
and are unknown variables. Supposing guide lines
and , and was devided into two angles, and
, by . Same as , it is devided into and
by . From geometry knowledge, and are
motor desired angles. Point P position is known variable.
Supposing guide lines and , The lengths of
and are provided in 11 12 . The cosine values of Fig. 4: DC motor sketch
, are shown in 13 14 .
|PA | x r y (11) For DC motor, reference [11] derived the following
transformation function.
|PA | x r y (12) DC motor relative parameters are shown table 1.
Parameters Amount Units
| |
cos θ |
(13) n N. m/A
| 253
L 121 μH

cos θ | |
(14) J 98.6 gcm
R 0.346 Ω
So from figure 2, motor one’s desired angle can be b 3 10 Ns/m
m 0.00641 (1:156)
derived,
θ cos θ cos θ (15) Tab. 1: DC motor specifications

As a symmetric mechanism, and cosine

values are derived in 16 17 . Motor two's desired angle
value is in equation 18 .
| |
cos θ |
(16)
|

cos θ | |
(17)
Fig. 5: DC motor model
θ cos θ cos θ (18)
Bond Graph of Five-Bar Robot

3 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Karnopp and Rosenberg made multiple rigid body model. In this arm one (proximal arm) sub-model,
It gives out Eulerian Junction Structure (EJS) model [12]. In
this work, robot arm is a rigid body. This model is built based AG BG = 125mm (20)
on body fixed frame and center mass. In this paper, as five-bar
V V AG sinθ ω 0 (21)
workspace is XY coordinate, gravity influence is less than XZ
or YZ coordinate. So its center mass is assumed as 100 grams. V V AG cosθ ω 0 (22)
Through frame transformation matrix , variables in body
V V BG sinθ ω (23)
frame can be transformed to ground frame.
ω ω (19) V V BG cosθ ω (24)

Every arm is an EJS model. Its rotation referring ground

frame and it can be shown in matrix. Bond graph is easy
to show the relationship between velocity and angular velocity.
It is a suitable method to model planar manipulator.
In this project, all variables are referred to the same ground
coordinate.
The bond graph sub-model of arm one is shown in figure 6.
In this model, J_G1 is an continuous power storage element and
represents arm rotation inertia referring to Z axis. Another two
M1x and M1y stand for beam mass. Since this mechanism
workspace is XOY Cartesian coordinate, 100 grams are
manually valued to this two elements. Same as other arms.
Where is the x direction velocity of arm one on point
, and is the y direction velocity of arm one on point
. Where is the x direction velocity of arm one on point
, and is the y direction velocity of arm one on point
. is the x direction velocity of arm one on mass center
G. is the y direction velocity of arm one on mass center
G. For the adoption of symmetrical configuration, arm two
's model can be built same method. Point and point Fig. 7: Bond graph of arm three B P
are fixed on ground, thus equations (19)(20) both equal to
zero. Arm three (distal arm) is EJS model and it is shown
in figure 7. In this sub-model, Where is the x direction
velocity of arm one on point , and is the y direction
velocity of arm one on point . Where is the x direction
velocity of arm one on point , and is the y direction
velocity of arm one on point . is the x direction velocity
of arm one on mass center G. is the y direction velocity
of arm one on mass center G. All these velocity variables are
referring to ground frame. For the adoption of symmetrical
configuration, arm four 's model can be built same
method.
PG BG 125mm (25)

V V BG sinθ ω (26)
V V BG cosθ ω (27)
V V PG sinθ ω (28)
V V PG cosθ ω (29)
Parasitic Element
It is relatively common to encounter derivative causality in
Fig. 6: Bond graph of arm one A B the mechanical part of the system due to the assumption that
inertia elements such as rigid bodies are connected rigidly [13].

4 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 To eliminate derivative causality, parasitic element is built to Arm Principle Units
Moments of inertia
isolate inertial components from the rest system. Figure 8 is the
2 ports parasitic element sub-model. It can be treated as a set of #1 (Proximal) 0.013562 ∙
parallel spring and damper with large values. In this model #2 (Proximal) 0.012068 ∙
#3 (Distal) 0.013556 ∙
and , which stand for damping
#4 (Distal) 0.012881 ∙
and compliance.
Tab. 2: Principle moments of inertia
Controller

Fig. 8: Bond graph isolator

4. SIMULATION RESULTS
In this section, simulation results are shown. As a whole
robotic system, controllers is studied. Dynamic results will be
shown in graphs. As figure 9, Integral bond graph model is built
in 20-sim. Parameters such as arms rotation inertia in 20-sim
are created in Solidworks Fig. 10: PID controller
The properties of this five-bar robot are needed. Referred
In 20-sim, PID controller structure is shown in figure 10.
to Z axis, four bars has four principle moments of inertia are
This controller has four parameters including proportional gain
shown in table 2.
Kp, derivative time constant Td, integral time constant Ti, and
tameness constant beta. In this model, two PID controllers are
implemented. They are used to control 24V DC motors. So in
integral model, after PID controller there are two limits ranging
from -24 to 24.
Trial and error method is implemented to gain robust
performance. PID Tuned parameters are offered in table 3.
Angle and position control results are in figure 11 and table 4.
Parameters Amount Units
Kp 2050 second
TauD 1970 second
Beta 0.95
TauI 2150 second
Tab. 3: PID Tuned Parameters (Same On Two Controllers)
Theta1 & Desired Theta1 X Position & Desired X Position

1.55 Desired_Theta_1 0.25 Desired_X_Position {meter}

Theta_1 {rad} X_Position {meter}
1.5 0.2

1.45 0.15

1.4 0.1

1.35 0.05

1.3 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 80
time {s} time {s}
Theta2 & Desired Theta2 Y Position & Desired Y Position
1.5 0.5
Desired_Theta_2 Desired_Y_Position {meter}
1 Theta_2 0.4 Y_Position {meter}
0.5 0.3
0
0.2
-0.5
0.1
-1
0
-1.5
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70
time {s} time {s}

Fig. 9: Bond graph of five-bar parallel robot Fig. 11: Angles and positions controlled results

5 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Theta 1 X YTheta 2 Arms Maximum Torque Minimum
(rad) Position Position
(rad) (N.m) Torque (N.m)
(meter) (meter) #1 (Proximal Arm) -0.38278 0.39580
Initial 1.57080 1.57080 0 0.498675 #2 (Proximal Arm) 1.18674 -1.16695
Final 1.31766 -1.42768 0.09960 -0.000475 #3 (Distal Arm) 0.00800 0.14354
Desired 1.31812 -1.42023 0.1 0 #4 (Distal Arm) 0.31577 -0.33532
Difference 0.00046 0.00745 0.00040 0.000475 Tab. 6: Motors max. and min. torque and angular velocity
Errors 0.182% 0.249% 0.4% 0.095% Arm1 Force Curve (Proximal) Arm2 Force Curve (Proximal)
Tab. 4: Angles and positions controlled results
1 Arm_1_Force_x {N} Arm_2_Force_x {N}
1
Typical point to point process, it needs two inputs, desired Arm_1_Force_y {N} Arm_2_Force_y {N}
0.5 0.5
X position and desired Y position (if this desired position is
reachable in - + configuration workspace). Then the algorithm 0 0
will automatically calculate out all parameters and variables.
-0.5 -0.5
Motors will drive this system to push end effector to set point.
-1 -1
Dynamics
DC Motor 1 & DC Motor 2 Torque Curves DC Motor 1 & DC Motor 2 Angular Velocity Curves 0 10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80
0 time {s} time {s}
DCMotor_1_Torque {N.m} DCMotor_1_Angular_Velocity {rad/s}
1
DCMotor_2_Torque {N.m} DCMotor_2_Angular_Velocity {rad/s} Arm3 Force Curve (Distal) Arm4 Force Curve (Distal)
2
-0.02 1 Arm_3_Force_x {N} Arm_4_Force_x {N}
Arm_3_Force_y {N} 1.5 Arm_4_Force_y {N}
0.5
1
0.5
0.5
-0.04

0 0
0
-0.5

-0.06 -0.5 -1
-0.5 -1.5
-1 -2
-0.08
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 90
time {s} time {s}
-1

Fig. 14: Force applied on proximal and distal arms

-0.1
0 2 4 6 8 10 0 10 20 30 40 50 60 70 80 90 100
time {s} time {s}
Arm Max Max Min Min Max Min
Fig. 12: Motors torque and angular velocity Force Force Force Force Force Force
X (N) Y (N) X (N) Y (N) (absol (absol
DC Maximum Maximum Minimum
Minimum ute) ute)
Motor Torque Angular Angular
Torque #1 1.19210 0.30232 -1.22862 -0.20836 1.2298 1.2462
(N.m) Velocity Velocity
(N.m) #2 1.34251 1.34251 -1.31690 -1.31690 1.8986 1.8624
(rad/s) (rad/s) 1.17007 -0.36887 -1.10840 0.28881 1.2268 1.1454
#3
#1 0.03772 -0.24814 0 -0.01691 2.10682 0.02677 -2.10074 0.13606 2.1070 2.1051
#4
#2 1.24112 -1.36744 0 -0.09695
Tab. 7: Force applied on proximal and distal arms
Tab. 5: Motors max. and min. torque and angular velocity
Arm1 & Arm2 Torque Curves (Proximal) Arm3 & Arm4 Torque Curves (Distal)
In figure 12 and table 5, motors torque and angular
Arm_1_Torque {N.m} 0.3 Arm_3_Torque {N.m} velocity output are important criteria for robot design. Peak
Arm_2_Torque {N.m} Arm_4_Torque {N.m}
1
torque is no more than 2 N.m, which is available.
0.2 In figure 13 and table 6, the torque changing during the
0.5
process are provided. In reality application, these plots can be
0.1
used to analyze the material properties and to select optimal
configuration.
0
0
In figure 14 and table 7, the force changing during the
-0.1
point to point process is shown. The absolute max. and min.
-0.5
values are calculated. These results can be used to optimize
-0.2 robot arm strength.
In order to optimize model objectiveness, 3D animation
-1
-0.3 was made to testify this whole process. The end effector can
0 2 4 6
time {s}
8 10 0 2 4 6
time {s}
8 10 successfully reach the setting point. Since the setting point is
within - + configuration workspace, during the whole moving
Fig. 13: Toque applied on arms

6 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 process, five-bar manipulator has not changed its original ACKNOWLEDGMENT
configuration. The Z in figure 15 is Z axis. Thanks to Dr. Rideout's course ENGI 9496 about bond graph
and his selfless help.

REFERENCES
[1] Bonev I.A. The true origins of parallel robots. January, 24, 2003,
http://www.parallemic.org/Reviews/Review007.html.
[2] Zhang K, Huang T, Wang C. Kinematics and dynamics analysis of a
planar hybrid five bar actuator[C]//Control, Automation, Robotics and
Vision, 2006. ICARCV'06. 9th International Conference on. IEEE, 2006:
1-6.
[3] Karnopp D, Margolis D. Analysis and simulation of planar mechanism
(a) (b) systems using bond graphs. J Mech Des, Trans ASME. 1979;101(2):187-
191.
[4] Paynter, Henry M., Analysis and design of engineering systems, The
M.I.T. Press, Boston, 1961 ISBN 0-262-16004-8.
[5] Karnopp, Dean C., Margolis, Donald L., Rosenberg, Ronald C.,
1990: System dynamics: a unified approach, Wiley, ISBN 0-471-62171-4.
[6] Campos L, Bourbonnais F, Bonev IA, Bigras P. Development of a five-
bar parallel robot with large workspace. . 2010;2:917-922.
[7] Jesus Cervantes-Sanchez J, Cesar Hernandez-Rodriguez J, Angeles J. On
the kinematic design of the 5R planar, symmetric manipulator.Mechanism
and Machine Theory. 2001;36(11-12):1301-
1313. http://dx.doi.org/10.1016/S0094-114X(01)00053-
(c) (d) 2.doi:10.1016/S0094-114X(01)00053-2.
Fig. 15: 3D animation [8] Sun S, Cheung JWF, Lou Y. A study on five-bar manipulators for
semiconductor packaging applications. . 2007:1811-1816.
[9] MacHo E, Altuzarra O, Pinto C, Hernandez A. Workspaces associated to
CONCLUSION assembly modes of the 5R planar parallel manipulator. Robotica.
This work shows bond graph is a good modeling tool to 2008;26(3):395-403.
[10] Zeid A. Bond graph modeling of planar mechanisms with realistic joint
simulate closed loop mechanisms. Velocity, force, torque and effects. Journal of Dynamic Systems, Measurement and Control,
angle velocity are all can be presented by bond graph model. Transactions of the ASME. 1989;111(1):15-23.
This work also demonstrates that bond graph model is not only [11] Karnopp D. Energetically consistent bond graph models in
able to solve dynamic problems but also kinematic problems of electromechanical energy conversion. Journal of the Franklin Institute.
1990;327(5):677-686. http://dx.doi.org/10.1016/0016-0032(90)90076-U.
planar manipulator. doi: 10.1016/0016-0032(90)90076-U.
As for the model application, it can be used to evaluate planar [12] D. Karnopp and R. C. Rosenberg, Analysis and Simulation of Multiport
Systems - The Bond Graph Approach to Physical System Dynamics. MIT
manipulator's workspace, to check material's durability, to Press, Cambridge, MA, 1968.
testify actuator's performance, to testify the feasibility of new [13] Karnopp D. Approach to derivative causality in bond graph models of
algorithm. mechanical systems. Journal of the Franklin Institute. 1992;329(1):65-
75.http://dx.doi.org/10.1016/0016-0032(92)90096-Y. doi: 10.1016/0016-
0032(92)90096-Y.

7 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use