Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

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

net/publication/262646868

Dynamics Modelling of Suspended Mobile Manipulators: An Explicit Approach


with Verification

Article · January 2011


DOI: 10.2316/Journal.205.2011.2.205-5300

CITATIONS READS
3 164

2 authors:

Mahdy Eslamy S.A.A. Moosavian


UoT K. N. Toosi University of Technology
23 PUBLICATIONS   248 CITATIONS    260 PUBLICATIONS   2,879 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

RoboWalk assistive device View project

Modeling and control of robotic systems equipped with flexible members View project

All content following this page was uploaded by S.A.A. Moosavian on 14 December 2014.

The user has requested enhancement of the downloaded file.


International Journal of Modelling and Simulation, Vol. 31, No. 2, 2011

DYNAMICS MODELLING OF SUSPENDED


MOBILE MANIPULATORS: AN EXPLICIT
APPROACH WITH VERIFICATION
Mahdy Eslamy∗ and Seyed Ali A. Moosavian∗

Abstract the base–manipulator interactions and the dynamics mod-


elling. However, on the other hand, we will have a system
Mobile platforms with one or more manipulators are of great interest with more flexibility and better performance even in unde-
for their wide area of applications and dexterity. Including a sirable situations such as moving on uneven environments
suspension system in such mobile platforms will certainly add to [1, 2]. The inclusion of suspension subsystem increases
their manoeuvrability and very considerably to their complexity. the safety of the moving vehicle as well as its mounted
From model-based control point of view, for such complicated robotic arms, in response to shocks and jerks transmitted to them
systems with suspension system, it is vital to have a concise set of due to irregularity of the open terrain. Consequently, a
dynamic equations of motion; thus, the focus of this paper is on suspended wheeled mobile robot (SWMR) can be consid-
the dynamics modelling of suspended mobile robots with multiple ered as a proper system for movement on rough and gentle
manipulators for control purposes. To derive dynamics equation uneven terrain [3, 4]. From control point of view, espe-
of motion for such complex systems, the Lagrange formulation is cially in case a model-based approach has to be adopted,
utilized. To do this, first, the system kinetic energy is obtained one should first consider deriving a proper dynamics mod-
and then the system potential and dissipated energy are detailed elling for such complicated systems. To the best of our
through vigorous mathematical analyses. Next detailed derivation of knowledge, a detailed explicit dynamics modelling for the
the system’s mass, non-linear velocity and stiffness/gravity matrices suspended mobile manipulators has not been suggested in
is discussed. As shown in the literature, the obtained dynamics the robotics literature which can be used for model-based
model is very useful for control purposes and dynamics analyses. control purposes, and hence this point forms the main
Finally to verify the dynamics modelling procedure, the equations body of discussion in this paper.
of motion are programmed in Maple and later used in Matlab and The dynamic interaction between a 1 degree-of-
compared with the dynamics analysis software, Msc.Adams. The freedom (DOF) manipulator and a suspended vehicle has
obtained results guarantee the accuracy of the obtained dynamics been discussed by Meghdari et al. [5]. The kinematics and
modelling. dynamics of a free-floating space manipulator system have
been described using the virtual manipulator approach by
Key Words Vafa and Dubowsky [6]. Papadopoulos and Dubowsky
[7] have employed a barycentric vector approach to study
Mobile manipulator, suspension system, explicit dynamics model, kinematics and dynamics of a single-arm space free-flying
verification and simulation robot (SFFR) in free-floating mode. Saha and Angeles ob-
tained a systematic method for the kinematics and dynam-
1. Introduction ics modelling of a 2 DOF automated guided vehicle (AGV)
[8, 9]. The Lagrange formulation and the idea of direct
Using their moving platforms, mobile robotic systems can path method (DPM) [10–12] have been utilized for deriv-
do tasks that may be out of reach of their manipula- ing the kinematics and dynamics of a space robotic system
tors as opposed to fixed-base robots. Therefore, it re- equipped with multiple arms in which the system kinetic
sults in more complicated dynamics couplings due to the energy is derived and divided into three formats which
simultaneous interactions between the moving base and describe the contribution of each part of kinetic energy to
the manipulators. Obviously, using a suspension system the final equations of motion. It has been shown that the
with the mobile platform will add to the complexity of DPM concept has fewer computations compared to other
∗Advanced Robotics & Automated Systems (ARAS) Labo-
criteria [12]. This method will be in focus in this paper to
ratory, Department of Mechanical Engineering, K.N. Toosi obtain dynamics of general terrestrial mobile manipulators
University of Technology, P.O. Box 19395-1999 Tehran, Iran; equipped with suspension systems, and also to investigate
e-mail: m.eslamy@gmail.com; moosavian@kntu.ac.ir the terms induced by considering the suspension effects on
Recommended by Prof. O.M. Al-Jarrah
(DOI: 10.2316/Journal.205.2011.2.205-5300) the mobile platform and the manipulators and finally to
112
is the vector of joint angles for the mth manipulator, as
described below:

Rb = (xG , yG , zG )T , Θb = (θy , θr , θp )T ,
(m) (m)
θm = (θ1 , . . . , θk )T (3)

where xG , yG , zG are the base CM position with respect


to the inertial XY Z coordinate axes and θy , θr , θp are the
(m)
base yaw, roll and pitch angles, respectively. Finally, θk
denotes the kth joint variable of the mth manipulator.
To analyse (1) and obtain the equations of motion of
the system, one has to derive the kinetic, potential and
Figure 1. The concept of DPM in suspended mobile ma- dissipated energy terms, which are detailed in the following
nipulator. sections.

3. Kinetic Energy
reach a concise set of mathematical calculations for such
challenging systems. Recently, the obtained dynamics To derive the terms of the total kinetic energy (T ) for the
model is used in the model-based object manipulation (m)
robotic system, it is divided into two parts Tb and Tk ,
control algorithm for a suspended mobile robotic system which are the kinetic energy for the suspended moving
[13]. The iterative Newton–Euler method is also used to platform and kth link of the mth manipulator, respectively.
extract dynamics modelling of mobile manipulators [14,
15]. The focus of this paper is on obtaining the dynamics 3.1 Kinetic Energy of the Moving Platform ( Tb )
equations of suspended mobile manipulators based on
the DPM using some simply defined vectors. Derivation To obtain Tb one can write:
of the equations of motion results in explicit expressions of
the system’s mass, non-linear velocity and gravity/stiffness 
1 ˙ p · R
˙ p dM
matrices and generalized forces. Unlike recursive dynamics Tb = R (4)
2
formulations, the obtained dynamics model is very useful
to be used for control algorithms for mobile manipulators.
Finally, the obtained explicit dynamics model of a general where R˙ p is the position vector of an arbitrary point “p”
suspended mobile manipulator, to appear for the first on the platform and is defined as:
time here, is programmed in Maple and the results are
used in Matlab to get verified with dynamical analysis R ˙ b + ω
˙ p = R  b × rp/c (5)
software Msc.Adams. The results guarantee the accuracy
of the procedures used for obtaining the suspended mobile where ω b is the moving platform angular velocity and
manipulator dynamics equations. rp/c is the vector from platform CM to the point “p”.
Substituting (5) in (4), one will finally obtain:
2. Fundamental Definitions and Preliminaries
1 ˙ b + 1 ω
˙ b · R
Tb = Mb R  T · Ib · ω
b (6)
To express the dynamics of the suspended mobile manipu- 2 2 b
lator system as depicted in Fig. 1, the Lagrange approach is
used together with DPM concept [11]. To derive equations in which Mb and Ib are the mobile platform mass and
of motion for such a complex system, one can write: moment of inertia, respectively.
 
d ∂T ∂T ∂U ∂UF 3.2 Kinetic Energy of the k th Link of the mth
− + + = Qi i = 1, . . . , N (1) (m)
Manipulator (T k )
dt ∂ q̇i ∂qi ∂qi ∂ q̇i

where T , U , and UF are the robotic system’s total kinetic, The same procedure can be used to obtain the kinetic
potential and dissipated (due to damping equipments) energy of the manipulators’ links, taking the point “p,”
energy. N describes the system DOF, qi , q̇i , and Qi are the this time, on the kth link of the mth manipulator, one can
ith element of the vector of the generalized coordinates, write:
speeds and forces, respectively, as defined below:
R ˙ b + r˙ (m) + ω
˙ (m) = R (m) (m)
 k × rp/c k (7)
pk k
 
T T
q = RbT , ΘTb , θ1T , θ2T , . . . , θm (2)
(m)
in which rk is the CM position vector of the kth link
Rb and Θb describe position vector of the suspended plat- of mth manipulator with respect to the platform CM and
(m)
form centre of mass (CM) and its Euler angles and θm rp/c k is the position of point “p” with respect to the CM
113
(m)
of the corresponding link and ω
k is the corresponding
angular velocity as below:
k

(m) (m) (m)
wk = w0 + θ̇i zi m = 1, . . . , n (8)
i=1

(m)
where zi is a unit vector along the axis of rotation of
(m)
the ith joint of the mth manipulator, and q̇i is the
corresponding joint angle rate. Also note that,
k−1

(m) O m + (m) (m) (m)
r k =R (R i − Li ) − Lk m = 1, . . . , n
i=1
(9)
as shown in Fig. 1. R O m is the vector form the platform
CM to the installing point of the mth manipulator. As Figure 2. A simplified model of the suspended platform.
(m) (m)
shown, Ri and L i are the necessary right/left vectors
for reaching the CM of the kth link of the mth manipulator. To obtain Us i , for springs potential energy considera-
It should be noted that, however, the necessary rotation tions, a simplified model of the suspended mobile platform
matrices for each body has to be taken into account while without the manipulators is depicted in Fig. 2. For the ith
obtaining these relations. Substituting (7) into (4) it is spring upper connection point, R  i , one can write:
obtained as:
Nm
n  Nm
n  i = R
R b + L
i i = 1, . . . , 4 (14)
 1  (m) ˙ 1
Tlinks =
(m)
Tk = Mi R ˙
b · Rb +  b · Ib · ω
ω b
m=1 k=1
2 m=1 i=1 2  i is the vector from platform CM to the ith
where L
1
n Nm  connection point. Assuming δi to be the corresponding
(m) (m) (m) (m) (m) (m)
+ Mi r˙ci · r˙ci + ω  i · Ii · ω
i displacement of the ith spring, one can obtain:
2 m=1 i=1

n N
˙  m
(m) (m)
δi = l◦ k − {(R  i ) · k k − hi k}
b + L i = 1, . . . , 4 (15)
+ Rb · M · r˙ i ci (10)
m=1 i=1
where l◦ is the free length of the spring and hi is the
where n is the number of manipulators and Nm is the distance from lower end point to earth.
number of links for each manipulator. At last the system The springs potential energy is written as:
total kinetic energy is obtained as follows:
4
 4
1  
 Nm
n  Us i = Ki δ i · δ i (16)
T = Tb + Tk
(m)
(11) i=1
2 i=1
m=1 k=1
where Ki is the ith spring stiffness. At this point, all three
It is seen that all necessary computations could be calcu- parts of (12) are obtained as discussed.
lated based on simply defined vectors.
5. Dissipated Energy
4. Potential Energy
To obtain the last required part of (1), i.e., the dissipated
The potential energy of the robotic system is divided into energy due to damping systems, the concept of Rayleigh’s
three parts: dissipation function is utilized, therefore it could be writ-
Nm
n  4 ten as:
 (m)

U = Ug b + Ug k + Us i (12)
1 ˙ ˙
m=1 k=1 i=1 UF i = Ci δi · δi (17)
2
(m)
where Ug b and Ug k are the gravitational potential energy
where UF i is the dissipated energy due to ith damping
of the moving platform and the kth link of the mth ma- ˙
nipulator, and Us i is the ith spring potential energy in arrangement, δi and Ci are the rate of the corresponding
ith suspension arrangement according to the platform four displacement with time and damping coefficient. Based on
˙
corners which are obtained as: (15), δi is obtained as:

 Nm
n  Nm
n 

 b, (m) (m) (m)
 b +r ) δ˙i = −{(R
˙ b + L
˙ i ) · k k − ḣik} (18)
Ug b =Mb g · R Ug k = Mk g · (R k
m=1 k=1 m=1 k=1
(13) one should note that, [17],
114
and hence,
˙ i = ω
L i
b × L (19)  N N

∂ ˙ ˙ ∂  ∂δ1  ∂δ1
Therefore, the total dissipation energy will be: (δ 1 · δ 1 ) = q̇k · q̇s (25)
∂ · qi ∂ q̇i ∂qk s=1
∂qs
k=1
4
 1 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙
UF = UF i = (C1δ1 · δ1 +C2δ2 · δ2 +C3δ3 · δ3 +C4δ4 · δ4 ) If we rewrite the right-hand side of (25), we will have:
i=1
2
(20)
N
 N
To this end, all three main components of (1) are obtained; ∂δ1 ∂δ1 ∂δ1 ∂δ1 ∂δ1 ∂δ1
q̇k · q̇s = q̇1 · q̇1 + q̇1 · q̇2
in the next section, the derivation procedure will be applied ∂qk ∂qs ∂q1 ∂q1 ∂q1 ∂q2
k=1 s=1
on these terms.
∂δ1 ∂δ1 ∂δ1 ∂δ1 ∂δ1 ∂δ1
+ q̇1 · q̇3 + q̇1 · q̇4 . . . q̇2 . q̇1
6. Mathematical Preliminaries and Derivation ∂q1 ∂q3 ∂q1 ∂q4 ∂q2 ∂q1
Procedure ∂δ1 ∂δ1 ∂δ1 ∂δ1 ∂δ1 ∂δ1
+ q̇2 · q̇2 + q̇2 · q̇3 + q̇2 · q̇4 + . . .
∂q2 ∂q2 ∂q2 ∂q3 ∂q2 ∂q4
In this section, the mathematical steps yielding the final
(26)
form of the system dynamics equations are discussed. A
detailed discussion on the derivation process for the kinetic
energy, T , of a space robotic system is comprehensively It can be noted that (26) can have a concise form as:
presented in [12], therefore we skip this term and just

use the results obtained for T in [12] in the following N
 N N 
 N
∂δ1 ∂δ1 ∂δ1 ∂δ1
sections. The terms induced by U and UF are detailed in q̇k · q̇s = · q̇k q̇s (27)
∂qk s=1
∂qs s=1
∂qk ∂qs
the following subsections. k=1 k=1

6.1 Derivation on Potential Energy Terms ˙


One can write for δ1 :
In this subsection, we consider the terms induced by   N N

∂U /∂qi . Applying the derivation on the terms of the ∂UF 1 1 ∂   ∂δ1 ∂δ1
= C1 · q̇k q̇s
potential energy, (12), and noting that, ∂ q̇i 2 ∂ q̇i ∂qk ∂qs
k=1 s=1
⎡ ⎤
∂   ∂δ1  ∂δ1 ∂δ1
(δ1 · δ1 ) = · δ1 + δ1 · = 2δ1 · (21) ⎢ q̇1 ⎥
∂qi ∂qi ∂qi ∂qi  ⎢ . ⎥
∂δ1 ∂δ1 ⎢
. ⎥
1 ∂δ1 ∂δ1 ⎢. ⎥
one can obtain: = C1 2 · ...2 · ⎢ . ⎥ (28)
2 ∂qi ∂q1 ∂qi ∂qN ⎢ . ⎥
⎢. ⎥
⎣ ⎦
∂U ∂δ1 ∂δ2 ∂δ3 ∂δ4
= K1δ1 · + K2δ2 · + K3δ3 · + K4δ4 · q̇N
∂qi ∂qi ∂qi ∂qi ∂qi
n Nm



∂ Rb  (m)

∂ Rb
(m)
∂rk
+ mb g · + g · Mk + Doing the necessary simplifications, we will have:
∂qi m=1
∂qi ∂qi
k=1
⎡ ⎤
(22)
⎢ q̇1 ⎥
   ⎢ . ⎥
∂δ1 ∂δ1 ⎢ ⎥
It should be noted that (22) has no q̇ or q̈ coefficient and .
∂UF 1 ∂δ1 ∂δ1 ⎢. ⎥
itself forms the gravity/stiffness matrix as will be detailed = C1 · ... · ⎢. ⎥ (29)
∂ q̇i ∂qi ∂q1 ∂qi ∂qN ⎢ . ⎥
in the following sections. ⎢. ⎥
⎣ ⎦
q̇N
6.2 Derivation on Dissipated Energy Terms

Here we consider the terms derived from ∂UF /∂ q̇i . Noting Noting that (28) and (29) are done just for the first
that, suspension equipment, one can conclude based on (23) that,

∂UF 1 ∂ ˙ ˙ ∂ ˙ ˙
= C1 (δ1 · δ1 ) + C2 (δ 2 · δ 2 ) ∂UF
∂ q̇i 2 ∂ q̇i ∂ q̇i
∂ qi
⎡ ⎤
∂ ˙ ˙ ∂ ˙ ˙ δ1 ∂ 
∂ δ1 ∂
δ2 ∂  δ2
+ C3 (δ3 · δ3 ) + C4 (δ4 · δ4 ) (23) ⎢C1

·
∂qi ∂q1
+ C2 ·
∂qi ∂q1
+ ⎥

∂ q̇i ∂ q̇i ⎢



⎢ δ3 ∂ 
∂ δ3 ∂
δ4 ∂  δ4 ⎥
⎢C3 · + C4 · · · · ⎥
⎢ ∂q ∂q ∂q ∂q ⎥
˙ =⎢ i 1 i 1 ⎥ q
The displacement rate for δ1 can be written as: ⎢

⎢ · · · C1
∂ 
δ1
·
∂ 
δ1
+ C2
∂ 
δ2
·
∂ 
δ2


+⎥
⎢ ∂qi ∂qN ∂qi ∂qN ⎥
⎢ ⎥
⎢ ⎥
N ⎣ ∂
δ3 ∂
δ3 ∂
δ4 ∂
δ4 ⎦
 ∂δ1 C3 · + C4 ·
δ˙1 = q̇s (24) ∂qi ∂qN ∂qi ∂qN
∂qs (30)
s=1
115
7. Vector of Generalized Forces where C1 is the set of all q̇ coefficients and C2 is dependent
on the non-linear velocity terms but has no coefficients:
The vector of generalized forces Q can be divided for those
applied by the actuators of the manipulators and those
N
b
∂R  ∂2R b ∂ωb ∂ωb
applied on the platform, if all external forces except the C1 ij =M · + · Ib ·
ones applied on the platform are zero, one can write: ∂qi s=1
∂q s ∂q j ∂ q̇ i ∂q j

n N

N

∂2ω b ∂Rb   m  ∂ 2r(m)
⎧ ⎫ ⎧ ⎫ +ω b · Ib · + · Mk
(m) ck
q̇s
⎨0 ⎬ ⎨F ⎬ ∂ q̇i ∂qj ∂qi m=1 ∂qs ∂qj
6×1 b k=1 s=1
Q= + JTb (31)
N n N

⎩τ ⎭ ⎩τ ⎭  ∂2R  (m)  m (m)
K×1 b b (m) ∂ rc k
+ q̇s · Mk
s=1
∂qs ∂qj m=1 k=1
∂qi
n  Nm


N

Assuming that Fb and τb are applied at the platform  (m) ∂
(m)
rc k  (m)
∂ 2rc k ∂ωk
(m)

CM, Jb is defined as: + Mk · q̇s +


m=1 k=1
∂qi s=1
∂qs ∂qj ∂ q̇i
(m)

⎧ ⎫ 2 (m)
(m) ∂ ωk (m) (m) ∂ ω k
⎨Ṙ ⎬ · Ik · +ω  k · Ik ·
b ∂qj ∂ q̇i ∂qj
= Jb q̇ (32)

⎩ ω̇ ⎭
b ∂δ1 ∂δ1 ∂δ2 ∂δ2 ∂δ3 ∂δ3
+ C1 · + C2 · + C3 ·
∂qi ∂qj ∂qi ∂qj ∂qi ∂qj

In which Jb is the Jacobian matrix which relates the ∂δ4 ∂δ4
platform linear and angular velocities to the vector of gen- + C4 · (36)
∂qi ∂qj
eralized speeds. To this end, the derivation of the dynamics
model for suspended mobile manipulator is completed and
we are about to finalize our approach. It can be seen that the dissipated energy-related terms
are joined with C1 . C2 is an N × 1 vector as:

8. Finalizing Dynamics Equations


n Nm

∂
ωb  (m) (m) ∂
ωk
(m)
C2i =− ω
 b · Ib · +  k · Ik ·
ω
Exploiting Lagrange equations, (1), and substituting the ∂qi m=1
∂qi
k=1
system kinetic, potential and dissipated energy terms, the (37)
dynamics model is obtained as:
Finally, the potential energy-dependent terms or, in
other words, gravity/stiffness matrix, G, is obtained in
H(q) q̈ + C(q̇ , q) + G(q) = Q (33) concise form as:

∂δ1 ∂δ2 ∂δ3 ∂δ4


The mass matrix, H, an N × N matrix which is com- Gi = K1 δ1 . + K2 δ2 . + K3 δ3 . + K4 δ4 .
∂qi ∂qi ∂qi ∂qi
prised of all q̈ coefficients, is written as:
b n N
 m (m)
∂R (m) ∂  rc k
+ M g . + g · Mk (38)
 b ∂Rb ∂qi m=1
∂qi
∂R ∂ωb ∂ωb k=1
Hij = M · + · Ib ·
∂qi ∂qj ∂ q̇i ∂ q̇j
n N

In the next section, the obtained dynamics equations
 m
(m) ∂
(m)
rc k ∂rc k
(m)
∂
ωk
(m)
(m) ∂
ωk
(m)
+ Mk . + .Ik . are first programmed in Maple and the results are used in
m=1 k=1
∂qi ∂qj ∂ q̇i ∂ q̇j Matlab, then validated with dynamical analysis software

n N
 m (m) b Msc.Adams.
(m) ∂
rc k ∂R
+ Mk ·
m=1 k=1
∂qi ∂qj 9. Verification, Results, and Discussion

n N
 m
(m) ∂
(m)
rc k ∂Rb
+ Mk . (34) To validate the procedure for obtaining the dynamics
∂qj ∂qi
m=1 k=1 equation, various validation scenarios (such as suspended
planar/spatial model with/without multiple manipulators
and different road profiles) have been done in which some
where M is the total mass of the system. Without losing
results are depicted in the following.
any generality, the non-linear velocity matrix, C, can be
As shown in Fig. 3, a spatial suspended 6-DOF mobile
divided into two parts:
platform with two 3-DOF manipulators was one of our
validation scenarios (the attached manipulators have the
configuration of PUMA 560 as shown in Fig. 3 in which
C = C1 q̇ + C2 (35) each has 3 DOFs (with no wrist)). Our steps to write
116
Figure 3. The suspended mobile manipulator modelled in Msc.Adams.

the suspended mobile manipulator dynamics equations in Table 1


Maple were as below: The Surface Profiles
1. Determining the system general configuration, i.e.,
number of arms, links, generalized coordinates and the i ai (m) λi (m)
related geometry, mass/inertia properties of all compo- 1 0.09 0.6
nents, and if needed positions of applied forces/torques
on the suspended platform/links. 2 0.08 0.6
2. Preliminary calculations: transformation matrices for 3 0.12 0.6
each link, rotation matrix between global and Euler
coordinates of suspended platform. 4 0.16 0.6
3. Kinematics calculations: CM position vectors and
velocities of all links, absolute angular velocity of all
links/platform. Table 2
4. Differentiations and calculation of partial derivatives. Properties of Each Manipulator Links
5. Finalizing calculations, i.e., determining H-matrix us-
(m) (m)
ing (34), C1 using (36), C2 using (37), G using (38), kth link Length (m) Mk (kg) Ik (kg m2 )
generalized forces using (31).
1 1 80 [8;8;0]
6. Obtaining dynamics equations of the suspended mobile
manipulators in the form of: H q̈ + C1 q̇ + C2 + G = Q 2 1 50 [0;5;5]
The surface profile was chosen to be sinusoidal wave in
the form of hi = ai sin(2π/λi (vt ± 1)) for the front and rear
wheels, where v is the vehicle CM velocity. The suspended correspondence. It has to be mentioned that we have also
platform specifications and those of the manipulators and done the verification for the case when the joint variables
other necessary information are given in Tables 1–3, and vary with time in quintic equation, and ultimately, in both
the figure of the suspended mobile manipulator modelled cases the results from Msc.Adams and those programmed
in Msc.Adams is seen in Fig. 3. in Maple/Matlab had very good correspondence with
Each manipulator joint’s trajectory is defined to be each other. We have used this modelling procedure for
(1) (1) (1)
θ1 = 0 θ2 = 0.15 sin(8t)θ3 = 0.2 sin(4t), which are the a 18-DOF mobile robotic system with two PUMA-type
same for both manipulators. manipulators to control and carry an object in a mixed
The obtained results from the verification experiments circular-straight path. As the results were shown, even for
are detailed in Figs 4–6. such a complicated system, the dynamics of the robot is
As it is seen, the results obtained from both the much less challenging for the necessary controller compu-
proposed approach and Msc.Adams have very good tations because of the method discussed earlier [16].
117
Table 3
Properties of Platform

Mass (kg) I (kg m2 ) Length (m) Width (m) Thickness (m) K (kN/m) C (kNs/m) Platform CM Velocity (m/s)
3,000 [270;1050;1300] 2 1 0.2 180 9 2

Figure 4. Variation of the platform pitch angle versus time obtained from Msc.Adams/proposed approach.

Figure 5. Variation of the platform roll angle versus time obtained from Msc.Adams/proposed approach.

Figure 6. Variation of the platform vertical pos. versus time obtained from Msc.Adams/proposed approach.

118
10. Conclusions [12] S.A.A. Moosavian & E. Papadopoulos, Explicit dynamics of free
flyers with multiple manipulator via SPACEMAPLE, Journal
of Advanced Robotics, 18 (2), 2004, 223–244.
In this paper, using Lagrange formulation, the equations [13] M. Eslamy & S.A.A. Moosavian, Control of suspended wheeled
of motion for a general terrestrial mobile manipulator mobile robots with multiple arms during object manipula-
equipped with suspension system were derived based on tion tasks, IEEE International Conference on Robotics and
DPM which is very useful for dynamics analyses, design Automation (ICRA), Japan, 2009, 3730–3735.
[14] S.A.A. Moosavian & K. Alipour, Moment-height tip-over mea-
studies and the development of model-based control algo- sure for stability analysis of mobile robotic systems, Proc. of
rithms for such complex systems. To do this, first, the the IEEE/RSJ International Conference on Intelligent Robots
system kinetic energy is obtained based on some simply and Systems (IROS), Beijing, China, 2006, 5546–5551.
defined vectors. Next the system potential and dissi- [15] S.A.A. Moosavian & K. Alipour, On the dynamic tip-over
stability of wheeled mobile manipulators, International Journal
pated energy are detailed through very astute mathemat- of Robotics and Automation, 22 (4), 2007, 322–328.
ical analyses. Next the derivation of the system’s mass, [16] M. Eslamy & S.A.A. Moosavian, Dynamics and cooperative
non-linear velocity and stiffness/gravity matrices is de- object manipulation control of suspended mobile manipulators,
Journal of Intelligent and Robotic Systems, 60 (2), 2010, 181–
tailed. As shown in the literature, the obtained dynam- 199.
ics model is very useful for control purposes. Finally to [17] L. Meirovitch, Methods of analytical dynamics (New York:
validate the proposed dynamics modelling procedure, the McGraw-Hill, 1970).
equations of motion are programmed in Maple and then
in Matlab and compared with the dynamics analysis soft- Biographies
ware, Msc.Adams. The obtained results depicted the accu-
racy of the obtained dynamics modelling and the proposed Mahdy Eslamy received his B.Sc.
approach. degree from Iran University of Sci-
ence and Technology (IUST) in
References 2005 and the M.Sc. degree from
K.N. Toosi University of Technol-
[1] S. Dubowsky & E.E. Vance, Planning mobile manipulator ogy in 2008, both in mechanical
motions considering vehicle dynamic stability constraints ma-
engineering, in Tehran. He is also
nipulators, Proc. of the IEEE Int. Conf. on Robotics and
Automation, Scottsdale, AZ, USA, 1989, 1271–1276. a member of ARAS laboratory in
[2] R. Siegwart & I.R. Nourbakhsh, Introduction to autonomous KNTU. His fields of research are
mobile robots (Delhi: Prentice-Hall, 2005). about mechanical design, control
[3] R.F. Abo-Shanab, N. Sepehri, & Q. Wu, On dynamic modeling
of robot manipulators: The method of virtual links, Proc. of
and robotics.
ASME Design Engineering Technical Conference, Montreal,
Canada, 2002, Paper #DETC’02/MECH-34225. Seyed Ali A. Moosavian pro-
[4] R.F. Abo-Shanab & N. Sepehri, Effect of base compliance on
the dynamic stability of mobile manipulators, Robotica, 20, fessor of mechanical engineering,
2002, 607–613. received his Ph.D. degree in Me-
[5] M. Meghdari, M. Durali, & D. Naderi, Investigating dynamic chanical Engineering from McGill
interaction between the one D.O.F manipulator and vehicle
of a mobile manipulator, Journal of Intelligent and Robotic
University in 1996. He has been
Systems, 28, 2000, 277–290. with the Mechanical Engineer-
[6] Z. Vafa & S. Dubowsky, On the dynamics of manipulators ing Department at K.N.T. Uni-
in space using the virtual manipulator approach, Proc. of the versity of Technology (Tehran)
IEEE International Conference on Robotics and Automation,
Raleigh, NC, 1987, 579–585.
since 1997. He teaches courses in
[7] E. Papadopoulos & S. Dubowsky, On the nature of control the areas of robotics, dynamics,
algorithms for free-floating space manipulators, IEEE Trans- automatic control, analysis and
actions on Robotics and Automation, 7 (6), 1991, 750–758. synthesis of mechanisms. His re-
[8] S.K. Saha & J. Angeles, Dynamics of nonholonomic mechanical
systems using a natural orthogonal complement, ASME Journal search interests are in the areas of dynamics modelling,
of Applied Mechanics, 58, 1991 238–244. motion/impedance control of terrestrial and space robotic
[9] S.K. Saha & J. Angeles, Kinematics and dynamics of a three- systems. He is one of the founders of the ARAS Research
wheeled 2-DOF AGV, Proc. of the IEEE Int. Conf. on Robotics
Center for Design, Manufacturing and Control of Robotic
and Automation, Piscataway, USA, 1989, 1572–1577.
[10] S.A.A. Moosavian & E. Papadopoulos, Dynamics and control Systems and Automatic Machineries.
of space free-flyers with multiple arms, Journal of Advanced
Robotics, 9 (6), 1995, 603–624.
[11] S.A.A. Moosavian & E. Papadopoulos, On the kinematics
of multiple manipulator space free-flyers, Journal of Robotic
Systems, 15 (4), 1998, 207–216.

119

View publication stats

You might also like