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

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 65, NO.

7, JULY 2018 5673

Magnetic Navigation System Utilizing a Closed


Magnetic Circuit to Maximize Magnetic Field
and a Mapping Method to Precisely Control
Magnetic Field in Real Time
Jaekwang Nam , Wonseo Lee , Student Member, IEEE, Eunsoo Jung,
and Gunhee Jang , Member, IEEE

Abstract—We propose a novel closed-circuit magnetic manipulation does not require an electric wire or battery, and the
navigation system (CMNS), which utilizes eight electromag- magnetic objects, thus, have the advantages of miniaturization
nets connected by back yokes to maximize a magnetic field. and long operating times. Due to these advantages, MNSs
We first show the effectiveness of a closed magnetic circuit
(CMC) and conduct a parametric analysis to design a sin- and magnetic devices have attracted a great deal of attention
gle CMC, which is utilized to construct the whole CMNS. in various fields, such as lab-on-a-chip, magnetic catheters,
A magnetic field mapping method is also developed utiliz- and magnetic robots [1], [2]. Magnetic catheters and robots
ing the finite-element method and polynomial regression are notable applications as they provide a minimally invasive
to evaluate and control the magnetic field over almost the
treatment regimen because they only need a small hole to be
whole workspace in real time. We investigated how the mag-
netic field changed based on the shape of core tips by com- inserted into the human body and they can perform various
paring the isotropic magnetic field control authority, which tasks, such as a diagnosis, treatment, and biopsy.
is the ability to generate an equal magnetic field over the Various MNSs have been developed to manipulate magnetic
workspace regardless of the position and direction of the robots and catheters [3]–[8]. There have been two types of MNSs
magnetic field. We experimentally verified the mathemati-
produced according to the type of magnets. The first type is MNS
cal assumption that the magnetic field generated from the
proposed CMNS can be linearly proportional to the applied made with permanent magnets. This type of MNS can generate a
current and that the magnetic field of the proposed CMNS relatively strong magnetic field without energy consumption [9].
can be expressed as a superposition of the magnetic fields However, the magnetic field cannot rapidly change because the
generated by each electromagnet. Finally, we verified the permanent magnet must be physically moved to change the field.
effectiveness of the developed CMNS by performing exper-
Because this motion requires a large amount of space, it is hard
iments related to steering a commercial magnetic catheter.
to increase the number of the magnets. Also, permanent magnets
Index Terms—Closed magnetic circuit (CMC), magnetic can cause safety problems because they cannot be turned OFF,
field mapping, magnetic navigation system. even in emergencies.
Several researchers have investigated these MNSs with per-
I. INTRODUCTION
manent magnets. The Niobe II by Stereotaxis is an MNS made
MAGNETIC navigation system (MNS) is a device that of two permanent magnets [3]. It was developed to steer the
A generates an external magnetic field to manipulate objects
made of magnetic materials. Under the magnetic field, these
catheter in the human body. Mahoney and Abbott also devel-
oped an MNS made of a single permanent magnet [4]. This
objects experience magnetic torque and force, which enable was developed for applications in stomach capsule endoscopy,
them to move and perform various tasks. This form of wireless and it utilizes a robotic arm to increase the degree of freedom
(DOF) of the motion of the capsule. 5-DOF manipulation was
Manuscript received August 22, 2017; revised October 30, 2017; demonstrated using a mockup magnetic capsule endoscope.
accepted November 20, 2017. Date of publication December 11, There is another type of MNS that utilizes electromagnets.
2017; date of current version March 6, 2018. This work was sup- Because this type of MNS utilizes electromagnets, the mag-
ported by the National Research Foundation of Korea under Grant
2015R1A2A1A05001837, funded by the Korean government (MSIP). netic field can be instantaneously controlled by adjusting the
(Corresponding author: Gunhee Jang.) applied current without any physical movement. In addition, it
The authors are with the Department of Mechanical Convergence En- is safer than the MNS with permanent magnets because it can
gineering and the Graduate School of Hanyang University, Hanyang
University, Seoul 04763, South Korea (e-mail: njk0651@naver.com; be turned OFF in an emergency. However, the magnetic field
justinleews@gmail.com; hackviper654@gmail.com; ghjang@hanyang. generated from this MNS is smaller than the field generated
ac.kr). from the MNS with permanent magnets even though electric
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. energy is required to generate a magnetic field [9]. Generating a
Digital Object Identifier 10.1109/TIE.2017.2782220 strong magnetic field is necessary, especially in biomedical ap-

0278-0046 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
5674 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 65, NO. 7, JULY 2018

Fig. 1. Magnetic flux density distribution around three sets of electromagnets in three cases. (a) Two electromagnets with only coils. (b) Two
electromagnets utilizing the magnetic cores inside the coils to amplify the magnetic field. (c) Two electromagnets utilizing the magnetic cores inside
the coils and a back-yoke to form the CMC. (The direction and amplitude of the magnetic flux density are displayed by colored arrows, and the
contour shows the amplitude of the magnetic flux density at the airgap.)

plications where magnetic devices need strong magnetic torque CMNS by connecting the back side of eight electromagnets to
or force to perform various tasks including unclogging, biopsy, the magnetic back-yoke as shown in Fig. 1(c). The CMC can
and drug-release. serve as a passage for the magnetic flux; therefore, it can pre-
For this reason, several researchers have investigated methods vent magnetic flux leakage and concentrate the magnetic flux
for increasing the magnetic fields produced by MNSs with elec- on the workspace. We also developed a mapping method for the
tromagnets. Bell et al. constructed an MNS with three pairs of magnetic field by utilizing finite-element analysis and polyno-
Helmholtz coils to generate a three-dimensional (3-D) magnetic mial regression to approximate and precisely control the mag-
field [5]. Choi et al. added a pair of Maxwell coils to the MNS to netic field of the workspace in real time. To verify the proposed
generate a magnetic field gradient. The Helmholtz and Maxwell CMNS, we constructed the CMNS and measured the magnetic
coils are simple conventional electromagnets used to generate field over the workspace. Finally, we demonstrated the effective-
magnetic torque and force, but their combination is structurally ness of the improved magnetic field by performing experiments
inefficient [10]. Jeon et al. developed saddle coils for use with to steer a commercial magnetic catheter.
the Helmholtz and Maxwell coils to provide a compact structure
[11]. These saddle coils can generate stronger magnetic fields II. DEVELOPMENT OF THE NOVEL CMNS
and magnetic field gradients than the previous ones within the
A. Parametric Analysis for a Single CMC
same structural scale. Nam et al. improved the MNS with saddle
coils by utilizing the resonance effect of the RLC circuit of the We first show the concept and the effectiveness of the pro-
coils to generate alternating magnetic fields with the same static posed CMC by comparing three sets of electromagnets as shown
magnetic field amplitude [12]. Several researchers investigated in Fig. 1. Fig. 1(a) shows the two electromagnets that only have
the method used to increase the magnetic field by utilizing a coils, which is a basic concept of the MNSs in [5], [11], and
soft magnetic core inside the coils. Kummer et al. developed [15]. Fig. 1(b) also shows the two electromagnets, but it has the
the OctoMag, which is composed of eight electromagnets with magnetic cores inside the coils to amplify the magnetic field.
a soft magnetic core [6]. The magnetic core can amplify the This system is called an open magnetic circuit, and it is a basic
magnetic field, and eight electromagnets can generate 5-DOF concept of the MNSs in [6]. Fig. 1(c) shows the two electromag-
motion in a magnetic robot inside the ocular vasculature. Al- nets utilizing the CMC. To form the CMC, the back sides of the
though the OctoMag can generate an amplified magnetic field magnetic cores are connected by the magnetic back-yoke. Be-
in the ocular vasculature environment, the arrangement of the cause the two electromagnets are connected by the back-yoke,
electromagnets is not suitable for the human body. Nguyen et al. the magnetic flux generated from the electromagnets can effec-
rearranged the eight electromagnets to accommodate the human tively pass through the magnetic back-yoke with a small mag-
body [7]. The rearranged MNS can guide a magnetic catheter netic leakage, and the magnetic flux can be focused on the air
through blood vessels. However, a stronger magnetic field is still gap. To compare the magnetic field generated from each electro-
desired to perform various tasks such as unclogging, biopsy, and magnet, we developed three finite-element models correspond-
drug-release in the pulsatile conditions of blood vessels. ing to Fig. 1(a), (b), and (c), which have 931 603, 1 557 334, and
To increase the magnetic field of the MNS, we propose the 1 558 651 tetrahedral elements, respectively. These elements are
closed-loop magnetic navigation system (CMNS) utilizing the enough to guarantee the convergence of the developed finite-
closed magnetic circuit (CMC), which has been utilized to var- element models. Each electromagnet has the same number of
ious electromagnetic machines to enhance their performance coil turns (1000 turns) and the same applied current (10 A).
[13], [14]. We first conducted a parametric analysis of a single The inner and outer diameters of the electromagnet are 180 and
CMC to maximize the magnetic flux density in the workspace. 200 mm, respectively, and the magnetic cores are not saturated
The designed single CMC is utilized to construct the whole for these values. We utilized pure iron (DT4C) as the magnetic
NAM et al.: MAGNETIC NAVIGATION SYSTEM UTILIZING A CLOSED MAGNETIC CIRCUIT 5675

be calculated as follows:
k
=N h
lcoil = 2πNw {Lcore + rcoil + 2rcoil (k − 1)}
k =1
 
2 (Lcore + rcoil ) + 2rcoil (Nh − 1)
= 2πNw Nh
2
= 2πNw Nh (Lcore + Nh rcoil ) (4)
2
Fig. 2. B-H curve of the pure iron (DT4C). Acoil = πrcoil (5)
where Nw , Nh , Lcore , and rcoil are the coil turns along the axial
and radial direction of the electromagnet, the radial length of
the core and radius of the coil, respectively. In (4), Nw and Nh
can be rewritten using rcoil as follows:
Wcoil
Nw = (6)
2rcoil
Lcoil
Nh = (7)
2rcoil
where Wcoil and Lcoil are the width and radial length of the
Fig. 3. CMC utilizing two electromagnets and its design variables. wound coil area, respectively. Utilizing (3)–(7), the resistance
of the coil can be calculated as follows:
(2Lcore + Lcoil ) Wcoil Lcoil ρcoil
core and back-yoke, and its B–H curve in Fig. 2 was utilized for R= 4 . (8)
the nonlinear finite-element analysis. The magnetic fields gen- 4rcoil
erated at the center of a workspace with a diameter of 300 mm We can also calculate total turns of coil (N) utilizing Nw and
were calculated to be 1.58, 18.56, and 43.26 mT, respectively. Nh as follows:
These results show that the CMNS can generate a stronger mag-
Wcoil Lcoil
netic field than the conventional MNSs with the same power N = Nw N h = 2 . (9)
consumption. 4rcoil
We conducted parametric analysis to determine the design Substituting (8) and (9) into (2), we can obtain the NI as
variables of a single CMC which maximizes the magnetic field follows:
in the workspace, while Fig. 3 shows the single CMC and its 
design variables. After this process, the proposed CMNS are 1 Pout Wcoil (L − Lcore )
NI = (10)
constructed by combining four designed CMCs in Section II-B. 2 (L + Lcore ) ρcoil
The generated magnetic field between the two electromagnets
can be modeled utilizing an ideal CMC as follows [16]: where L is the summation of Lcore and Lcoil . Finally, we can
obtain the magnetic field generated by the two electromagnets
2μ0 N I utilizing (1) and (10) as follows:
B= (1)
l 
where μ0 , N, I, and l are the magnetic permeability of the air, μ0 Pout Wcoil (L − Lcore )
B= (11)
the number of coil turns, the current of the coil, and the length l (L + Lcore ) ρcoil
of the air gap, respectively. Equation (1) can be used to obtain
In (11), the ideal model suggests that the maximum values of
the maximum magnetic field with the maximum NI while l is
Wcoil and Pout , and the minimum value of Lcore are required to
a fixed value that depends on the size of the workspace. In this
generate the maximum magnetic field while the other variables
equation, we introduce the output of the power supply (Pout ),
are dependent on these three variables. However, we should
and the resistance of the coil (R). Then, NI can be expressed as
consider the saturation and fringing effects, which cannot be
follows:
 considered in the ideal model.
Pout The two design variables (Wcoil and Pout ) have little effect
NI = N . (2)
R on the saturation and fringing effects of the proposed CMC.
Thus, we can determine these values based on the ideal CMC
The resistance in (2) can be written as follows: model. At first, we considered the maximum Wcoil to maximize
the magnetic field. This value is constrained by the space limits
lcoil
R= ρcoil (3) of the CMNS, so we selected Wcoil to be 300 mm, which is
Acoil the maximum value considering the space limit. The maximum
where lcoil , Acoil , and ρcoil are the length, cross sectional area, Pout is another desired value in the ideal CMC model, and this
and resistivity of the coil, respectively. In (3), lcoil and Acoil can value is limited by the capacity of the power supply. Fig. 4
5676 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 65, NO. 7, JULY 2018

When Lcore is smaller than 90 mm, the generated magnetic flux


density at the core exceeds the saturation magnetic flux density,
which is shown in Fig. 2. Because the magnetic core is already
saturated, the magnetic flux density at the air gap is mainly af-
fected by fringing effect. Thus, the magnetic flux density at the
air gap increases until it reached 90 mm, as shown in Fig. 5(b);
this is because the cross-sectional area of the core (πL2core ) is
inversely proportional to the fringing effect, which reduces the
magnetic flux density at the air gap [17]. This fringing effect has
greater influence than the change in NI with the saturated mag-
netic core. However, this relationship was reversed when the
Fig. 4. Output range of the power supply (CSW5550 by California
Instruments). magnetic core was not saturated. Thus, the magnetic flux den-
sity at the air gap decreased when Lcore is longer than 90 mm,
as shown in Fig. 5(b). Finally, we selected Lcore of 90 mm to
maximize the magnetic field. Utilizing the selected design vari-
ables of Wcoil (300 mm), Pout (1850 VA), and Lcore (90 mm),
we maximized the magnetic field at the air gap.

B. Development of CMNS
We placed four sets of the designed CMC 90° apart along the
circumference of a circle. Then, the CMNS has eight electro-
magnets, which are enough to generate 3-D magnetic torque and
force because they have eight components by Maxwell equation
[18]. In this arrangement, the eight ends of the electromagnets
form a regular hexahedron, and the upper and lower electro-
magnets have the tilting angles of 54.74° and −54.74° with
respect to the z-axis. We connected them with two cross-shaped
back-yokes to form a closed-circuit magnetic navigation system
while they have greater cross-sectional area (110 × 250 mm2 )
than the magnetic core (π × 902 mm2 ) to prevent magnetic
saturation. The height and width of the CMNS were 1860 and
1760 mm, respectively, and an inner space with a diameter of
300 mm was formed at the center of the CMNS.

C. Magnetic Field Mapping and Control Methodology


We can generate magnetic fields and field gradients utilizing
the developed CMNS. However, we focused on generating the
Fig. 5. Calculated magnetic flux density of the CMC by finite-element magnetic field because the magnetic torque is more effective
analysis according to L c o re (a) at the center of the magnetic core and than the magnetic force to manipulate the magnetic robots in
(b) at the center of the air gap.
subcentimeter scale [19]. Magnetic torque on a magnetic device
generated by an external magnetic field can be expressed as
follows [20]:
shows the output range of the power supply (CSW5550 by Cal-
ifornia Instruments). As shown in Fig. 4, the maximum Pout of T=m×B (12)
1850 VA can be generated in the resistance ranging from 7.19 to where m and B are the magnetic moment of a magnetic device
13.23 Ω due to the current and voltage limits of the power sup- and external magnetic field, respectively. We assumed that each
ply. The resistance in this range can be obtained by adjusting electromagnet generated an independent magnetic field and that
rcoil in (8). However, the design variable Lcore affects the sat- the total magnetic field generated by eight electromagnets can
uration and fringing effects because these effects are related to be superposed as follows [6]:
the cross sectional area of the magnetic core.
8

We conducted finite-element analysis utilizing the finite-
element model in Fig. 1(c) to consider both the saturation and B (P) = Bm (P) (13)
m =1
fringing effect. The maximum Pout and Wcoil were applied for
this calculation while the input NI changed according to Lcore where P and Bm are any given point (x, y, z) in the workspace
in (10). Fig. 5 shows the calculated magnetic flux density at and the magnetic field from the mth electromagnet at the point
the center of the magnetic core and the center of the air gap. P, respectively. We also assume that the magnetic field of each
NAM et al.: MAGNETIC NAVIGATION SYSTEM UTILIZING A CLOSED MAGNETIC CIRCUIT 5677

electromagnet is linearly proportional to the current in the case constants of the polynomial regression models, we utilized the
that the magnetic core and back-yoke are unsaturated. Then, calculated magnetic field at N sample points. Then, we obtained
Bm can be rewritten as follows: N simultaneous equations by applying the results at N sample
points in (18), (19), and (20) as follows:
Bm (P) = B̃m (P) im (14)
⎡ ⎤
where B̃m and im are the magnetic field generated by the mth ⎡ FE ⎤ ⎡ n n n ⎤ (am x )n 00
electromagnet by a unit current and the current applied to the mth Bm x 1 x1 y1 z1 · · · 1 ⎢ ⎥
⎢(am x )0n 0 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎦ = ⎣ .. .. .. . . . .. ⎦ ⎢(am x )00n ⎥
electromagnet, respectively. Using (14), (13) can be rewritten BFmEx = ⎣ .
. . . . .
 F .E  ⎢ . ⎥
as follows:
⎡ ⎤ Bm x N xnN yNn n
zN ··· 1 ⎣ .. ⎦
i1 1
 ⎢ . ⎥ = AF E am x
B (P) = B̃1 (P) · · · B̃8 (P) ⎣ .. ⎦ = B (P) I. (15)
(21)
i8
In this equation, we can generate the desired magnetic field ⎡ ⎤
at point P utilizing the current vector I, and this vector can be ⎡  ⎤ ⎡ ⎤ (am y )n 00
FE
Bm y 1 xn1 y1n z1n · · · 1 ⎢ ⎥
⎢(am y )0n 0 ⎥
calculated using the pseudoinverse B(P)† as follows [21]: ⎢ .. ⎥ ⎢ .. .. .. . . .. ⎥ ⎢(am y )00n ⎥
BFmEy =⎣ .  ⎦=⎣ . . . . .⎦ ⎢ ⎥
I = B(P)† B (P) .  ⎢ .. ⎥
(16) FE
Bm y N xnN n
yN zNn
··· 1 ⎣ . ⎦
We can generate the desired magnetic field at point P by 1
providing the currents in (16) to the eight electromagnets. = AF E am y
There are several methods to obtain B̃m in (14). One simple (22)
approach is to utilize mathematical models such as the magnetic
dipole model [6]. This model provides a continuous function of ⎡ ⎤
⎡  ⎤ ⎡ ⎤ (am z )n 00
the magnetic field and a relative exact solution by fitting the FE
Bm z 1
xn1 y1n z1n · · · 1 ⎢ ⎥
⎢(am z )0n 0 ⎥
model to the small workspace. However, this model cannot pro- ⎢ .. ⎥ ⎢ .. .. .. . . .. ⎥ ⎢(am z )00n ⎥
BFmEz =⎣ . ⎦=⎣ . . . . .⎦ ⎢ ⎥
vide an exact solution in a large workspace. Another approach   ⎢ .. ⎥
is to utilize the finite-element model. This model can provide an FE
Bm z N xnN n
yN zNn
··· 1 ⎣ . ⎦
exact solution even in the large workspace. However, it cannot 1
provide the solution in real time because long computational = AF E a m z
time is required to solve for a nonlinear magnetic field. There- (23)
fore, we developed a mathematical polynomial regression model where BFmEx , BFmEy , and BFmEz are the x-, y-, and z-directional
utilizing the results of the finite-element analysis. The devel- calculated magnetic fields at N sample points and AF E is the
oped polynomial regression model of B̃m can be expressed as matrix in terms of x, y, and z. In these equations, the size of matrix
follows: AF E is N × Nterm s while the number of terms (Nterm s ) can
  be calculated by the formula for the combination with repetition
B̃m = B̃m x B̃m y B̃m z (17)
as follows [22]:
=n j
k =n 
i=n
n

B̃m x (x, y, z) = (am x )ij k xi y j z k , (i+j +k ≤ n) Nterm s = 3 Hk . (24)
k =0 j =0 i=0
k =0
(18)
Finally, we can obtain the constants of the polynomial regres-
=n j
k =n 
i=n
sion models utilizing the pseudoinverse as follows [21]:
B̃m y (x, y, z) = (am y )ij k xi y j z k , (i+j +k ≤ n)
k =0 j =0 i=0
 †
am x = AF E BFmEx (25)
(19)
 †
=n j
k =n 
i=n am y = AF E BFmEy (26)
i j k  †
B̃m z (x, y, z) = (am z )ij k x y z , (i+j +k ≤ n)
k =0 j =0 i=0
am z = AF E BFmEz . (27)
(20) Using this method and 113 sample points, we obtained
where (am x )ij k , (am y )ij k , and (am z )ij k are the constants of the polynomial regression models with the average error of
the polynomial regression models of Bm x , Bm y , and Bm z . 1.3% while their magnetic flux densities were compared with
To consider the interactions between x, y, and z, the equations those calculated by the finite-element method at 113 sample
contain all product combinations under n order of x, y, and z. points. Because the developed polynomial regression models
In the equations, the value of n can change to reduce the error can quickly provide the continuous magnetic field map over the
with the finite-element model, and we determined n to be 5, large workspace, we can easily control the magnetic field in real
which matches well with the proposed CMNS. To obtain the time.
5678 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 65, NO. 7, JULY 2018

Fig. 8. (a) Workspace divided into eight regular cubes. (b) Cube with
512 sample points. (c) Magnetic field orientations expressed by 361 half
cardinal orientations at each point.

G
Using the above-mentioned method, we can calculate Bm in ,
Fig. 6. Proposed CMNS utilizing the eight electromagnets and
back-yokes. which represents the index of the isotropic magnetic field control
authority, according to the tips. Because the polynomial regres-
sion magnetic field model in (17) requires the results of the
finite-element analysis, we conducted the finite-element analy-
sis for the each core tip. Each finite-element model had about
600 000 tetrahedral elements, which are sufficient to guaran-
tee the convergence of the finite-element model. We assumed
that only 80% of the cubic inner space of the CMNS can be
utilized in actual operation. However, we only utilize a part of
workspace because the CMNS has a symmetrical arrangement.
Fig. 7. Various tips for the CMNS. (a) Convex tip. (b) Concave tip.
(c) Flat tip. (d) Square penetration tip.
We divided the workspace into eight cubes and utilized one
of them as shown in Fig. 8(a), while a divided cube has 512
sample points as shown in Fig. 8(b). Utilizing the results of
the finite-element analysis at 512 sample points, we obtained
D. Effect of Core Tip the polynomial regression model of B̃m in (17). We consider
The core tip is one of the important parts of the CMNS be- 361 half cardinal orientations of the magnetic field to calculated
L
cause the magnetic field is generated through the core tip, and the Bm in at each sample point utilizing the developed model as
L
magnetic field distribution over the workspace changes based on shown in Fig. 8(c). The calculated Bm in is displayed in color at
G
the shape of the core tip. We considered four core tip including each sample point as shown in Fig. 9. We observed that Bm in
convex, concave, flat, and square penetration types, as shown exists at the vertex of the cube in every core tip, and these values
in Fig. 7(a) to (d). The distance between two opposite electro- were 59.4, 58.2, 50.6, and 59.0 mT, respectively. Considering
magnets was fixed as 300 mm for all tips to secure the same these results, the convex tip may be the best solution to generate
size of the inner space. We developed a method to evaluate an the isotropic magnetic field. However, the penetration type can
isotropic magnetic field control authority, which is the ability to effectively accommodate a long subject such as human body.
generate an equal magnetic field over the workspace regardless In addition, the square penetration has almost the same value
G
of position and direction. Utilizing this method, we evaluated of Bm in with the convex tip. Therefore, we selected the square
the CMNS with different types of core tips. penetration tip for the proposed CMNS.
The first step to evaluate the authority is to calculate an avail-
able maximum magnetic field in a specific direction at a target III. RESULT AND DISCUSSION
point P. To calculate this value, we considered the current limit
of the power supply. In (16), the current vector can be calcu- A. Construction and Verification of the CMNS
lated to generate the desired magnetic field at a target point P. Fig. 10 shows the constructed CMNS and the experimental
However, the desired magnetic field cannot be generated even if setup used to generate the magnetic field. Because the construc-
one element of I reached the current limit of the power supply. tion process required mechanical polishing of the magnetic core,
Under this limitation, we can calculate the available maximum the designed value of Lcore (90 mm) was slightly reduced to
magnetic fields at a point P in specific direction. We defined 88.5 mm. However, this change did not significantly affect the
L
Bm in as the smallest magnetic field among the available mag- magnetic field as shown in Fig. 5(b). The width of the wound
netic fields in every direction at a point P, and we calculated coil area (Wcoil ) was the same as the designed 300 mm because
it on every sample point in the workspace to find the global we guided the coil using the coil holder. The coils were wound
G
minimum (Bm in ). Then, the global minimum can be an index with a radius (rcoil ) of 0.75 mm and 1333 turns resulting in a
of the isotropic magnetic field control authority because we can resistance of 7.19 Ω, which is in the range of the maximum
G
generate a greater magnetic field than Bm in at any point P and Pout as shown in Fig. 4. However, the measured resistance of
direction. the wound coils was greater than the simulated resistances as
NAM et al.: MAGNETIC NAVIGATION SYSTEM UTILIZING A CLOSED MAGNETIC CIRCUIT 5679

Fig. 11. Comparison of the magnetic field between the experiment


(B E X P ) and finite-element model (B F E ) at (a) nine experimental points.
The magnetic fields were compared in the (b) x-, (c) y-, and (d) z-direction
at the experimental points while a current of 10 A was applied to the
electromagnet (L1) in Fig. 6.
Fig. 9. Calculated B m L
in in a divided workspace according to the
various tips. (a) Convex tip. (b) Concave tip. (c) Flat tip. (d) Square
penetration tip.

Fig. 12. Magnetic field generated by U1 at the center of the workspace


according to the applied current.

Fig. 11, and we measured the magnetic field utilizing a Gauss


probe (Model 6010 by F. W. Bell) while a current of 10 A was
applied. Fig. 11(b) to (d) show the results of the electromagnet
Fig. 10. Constructed CMNS and experimental setup to generate the
L1 in Fig. 6, which has the maximum error among the eight elec-
magnetic field. tromagnets. This experiment shows that the constructed CMNS
generated the magnetic field, which matched well with the mag-
netic field of the finite-element model under a maximum error
TABLE I
RESISTANCES OF THE EIGHT ELECTROMAGNETS
of 5%.

Electromagnet U1 U2 U3 U4 L1 L2 L3 L4 B. Verification of the Linear and Independent Control


Resistance (Ω) 8.24 8.14 8.06 8.36 8.33 8.36 8.49 8.42 Capability
We assumed that the magnetic field of each electromagnet
was linearly proportional to the current in (14) under the con-
shown in Table I because the coils could not be ideally wound. dition that the magnetic core and back-yoke were not saturated
The maximum currents decreased by 5–8% due to this effect. by the applied current. To verify this assumption, we measured
Because the magnetic field is linearly proportional to the applied the magnetic field generated by an electromagnet (U1) as a
current, the maximum magnetic fields also decreased based on function of the current. Fig. 12 shows the measured magnetic
the current applied to the electromagnets. field at the center of the workspace, and we observed that three
We compared the measured magnetic field with the mag- components of the magnetic field linearly increased up to the
netic field of the finite-element model to verify the constructed maximum current of 15 A. We also assumed that the magnetic
CMNS. We selected nine experimental points as shown in field of the proposed CMNS was superposed by the magnetic
5680 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 65, NO. 7, JULY 2018

time. We first designed a single CMC by conducting a paramet-


ric analysis and utilized it to construct the CMNS with eight
electromagnets and back yokes. The mathematical magnetic-
circuit model was utilized for the parametric analysis, and the
finite-element model was also used to consider the magnetic sat-
uration and fringing effect. We also investigated the magnetic
field according to shapes of core tips by comparing the isotropic
magnetic field control authority. The developed CMNS could
generate a magnetic field of 59.0 mT regardless of the position
and direction inside the cubic workspace. We experimentally
verified the mathematical assumption that the magnetic field
generated from each electromagnet is linearly proportional to
the applied current and it can be superposed. Finally, we verified
the effectiveness of the developed CMNS by measuring steer-
ing angle of a commercial magnetic catheter. We expect that
the proposed CMNS can improve the performance of magnetic
Fig. 13. Steering motion of the 5-cm-long magnetic catheter based on catheters and magnetic robots in biomedical fields.
the amplitude of the magnetic field with (a) 14 mT, (b) 36 mT, (c) 109 mT,
and (d) angle difference between the magnetic field (θB ) and catheter
(θs ).
REFERENCES
field generated by each electromagnet as shown in (13). To [1] M. Hagiwara, T. Kawahara, Y. Yamanishi, T. Masuda, L. Feng, and F.
verify this assumption, we measured the two individual mag- Arai, “On-chip magnetically actuated robot with ultrasonic vibration for
netic fields generated from U1 and U2 based on a current of single cell manipulations,” Lab. Chip, vol. 11, no. 12, pp. 2049–2054,
2011.
10 A. The individual magnetic fields generated from U1 and [2] W. Lee, J. Nam, B. Jang, and G. Jang, “Selective motion control of a
U2 were measured to be [−14.92 mT, −14.11 mT, −14.54 mT] crawling magnetic robot system for wireless self-expandable stent delivery
and [14.91 mT, −14.31 mT, −15.24 mT]. The magnetic field in narrowed tubular environments,” IEEE Trans. Ind. Electron., vol. 64,
no. 2, pp. 1636–1644, Jun. 2016.
generated from both U1 and U2 was measured to be [0.00 mT, [3] C. Pappone et al., “Robotic magnetic navigation for atrial fibrillation
−28.42 mT, −29.78 mT]. This result shows that the magnetic ablation,” J. Amer. College Cardiol., vol. 47, no. 7, pp. 1390–1400, Apr.
field of the proposed CMNS can be superposed by the magnetic 2006.
[4] A. W. Mahoney and J. J. Abbott, “Five-degree-of-freedom manipulation
field generated by each electromagnet. of an untethered magnetic device in fluid using a single permanent mag-
net with application in stomach capsule endoscopy,” Int. J. Robot. Res.,
C. Steering of a Magnetic Catheter vol. 35, no. 1-3, pp. 129–147, Feb. 2015.
To demonstrate the effectiveness of the constructed CMNS, [5] D. J. Bell, S. Leutenegger, K. M. Hammar, L. X. Dong, and B. J. Nelson,
“Flagella-like propulsion for microrobots using a nanocoil and a rotating
we compared the steering angle of the magnetic catheter (Cel- rlectromagnetic field,” in Proc. 2007 IEEE Int. Conf. Robot. Autom., 2007,
sius ThermoCool RMT, Biosense Webster) as a function of pp. 1128–1133.
the magnetic field. We first placed the magnetic catheter along [6] M. P. Kummer, J. J. Abbott, B. E. Kratochvil, R. Borer, A. Sengul, and
B. J. Nelson, “OctoMag: An electromagnetic system for 5-DOF wireless
the x-axis at the center of the xy-plane and applied the mag- micromanipulation,” IEEE Trans. Robot., vol. 26, no. 6, pp. 1006–1017,
netic field along the x-axis. Then, we rotated the magnetic Dec. 2010.
field from 0° to 180° and measured the steering angle of [7] B. L. Nguyen, J. L. Merino, and E. S. Gang, “Remote navigation for
ablation procedures—A new step forward in the treatment of cardiac
the catheter. For this experiment, we applied three magnetic arrhythmias,” Eur. Cardiol. Rev., vol. 6, no. 3, pp. 50–56, 2010.
fields of 14, 36, and 109 mT, which represent the magnetic [8] L. Arcese, M. Fruchard, and A. Ferreira, “Adaptive controller and observer
fields of a conventional MNS with saddle coils [11], and the for a magnetic microrobot,” IEEE Trans. Robot., vol. 29, no. 4, pp. 1060–
1067, Aug. 2013.
proposed CMNS without and with a back-yoke, respectively. [9] S. Erni, S. Schürle, A. Fakhraee, B. E. Kratochvil, and B. J. Nelson,
Fig. 13 shows the experimental results. The angle difference “Comparison, optimization, and limitations of magnetic manipula-
between the magnetic field (θB ) and catheter (θs ) increased tion systems,” J. Micro-Bio Robot., vol. 8, no. 3/4, pp. 107–120,
Nov. 2013.
with increasing θB due to the stiffness of the catheter. However, [10] H. Choi, K. Cha, S. Jeong, J.-O. Park, and S. Park, “3-D lo-
the strong magnetic field can reduce this angle difference as comotive and drilling microrobot using novel stationary EMA sys-
shown in Fig. 13(d). Utilizing the proposed CMNS, we signif- tem,” IEEE/ASME Trans. Mechatron., vol. 18, no. 3, pp. 1221–1225,
Jun. 2013.
icantly reduced the angle difference to 20° in the case that θB [11] S. Jeon, G. Jang, H. Choi, and S. Park, “Magnetic navigation system
is 180°. This value is 3 and 6 times smaller than the CMNS with gradient and uniform saddle coils for the wireless manipulation of
without a back-yoke and the conventional MNS with saddle micro-robots in human blood vessels,” IEEE Trans. Magn., vol. 46, no. 6,
pp. 1943–1946, May 2010.
coils. [12] J. Nam, W. Lee, B. Jang, and G. Jang, “Magnetic navigation system
utilizing resonant effect to enhance magnetic field applied to magnetic
robots,” IEEE Trans. Ind. Electron., vol. 64, no. 6, pp. 4701–4709,
IV. CONCLUSION Feb. 2017.
We proposed a novel CMNS to maximize magnetic fields [13] M. Yamaguchi, S. Arakawa, H. Ohzeki, Y. Hayashi, and K. I. Arai,
“Characteristics and analysis of a thin film inductor with closed mag-
and proposed a magnetic field mapping method to evaluate and netic circuit structure,” IEEE Trans. Magn., vol. 28, no. 5, pp. 3015–3017,
control magnetic fields over the almost whole workspace in real Sep. 1992.
NAM et al.: MAGNETIC NAVIGATION SYSTEM UTILIZING A CLOSED MAGNETIC CIRCUIT 5681

[14] Z. Q. Zhu, Y. Pang, D. Howe, S. Iwasaki, R. Deodhar, and A. Pride, Eunsoo Jung received the B.S. degree in me-
“Analysis of electromagnetic performance of flux-switching permanent- chanical engineering in 2017 from Hanyang
magnet machines by nonlinear adaptive lumped parameter magnetic cir- University, Seoul, South Korea, where he is cur-
cuit model,” IEEE Trans. Magn., vol. 41, no. 11, pp. 4277–4287, Nov. rently working toward the Ph.D. degree in me-
2005. chanical convergence engineering.
[15] H. Choi et al., “EMA system with gradient and uniform saddle coils for His research interests include designing
3D locomotion of microrobot,” Sensors Actuators Phys., vol. 163, no. 1, magnetic microrobots and magnetic catheters
pp. 410–417, Sep. 2010. manipulated by magnetic navigation system in
[16] R. Merzouki, A. K. Samantaray, P. M. Pathak, and B. O. Bouamama, In- human blood vessels.
telligent Mechatronic Systems: Modeling, Control and Diagnosis. Berlin,
Germany: Springer-Verlag, 2012.
[17] R. Jez and A. Polit, “Influence of air-gap length and cross-section on
magnetic circuit parameters,” in Proc. COMSOL Conf., 2014, pp. 1–6.
[18] N. J. Groom, “Expanded equations for torque and force on a cylindrical
permanent magnet core in a large-gap magnetic suspension system,” NASA
Langley Res. Center, Hampton, VA, NASA Tech. Paper 3229, Feb. 1997.
[19] J. J. Abbott et al., “How should microrobots swim?” Int. J. Robot. Res., Gunhee Jang (M’00) received the B.S. de-
vol. 28, no. 11/12, pp. 1434–1447, Nov. 2009. gree from Hanyang University, Seoul, South Ko-
[20] L. Vaidman, “Torque and force on a magnetic dipole,” Am. J. Phys., vol. 58, rea, in 1984, the M.S. degree from the Ko-
no. 10, pp. 978–983, Oct. 1990. rea Advanced Institute of Science and Tech-
[21] R. A. Horn and C. R. Johnson, Matrix Analysis, 2 ed. New York, NY, nology, Seoul, South Korea, in 1986, and the
USA: Cambridge Univ. Press, 2012. Ph.D. degree from the University of California,
[22] D. P. Bertsekas and J. N. Tsitsiklis, Introduction to Probability, 2nd ed. Berkeley, CA, USA, in 1993, all in mechanical
Belmont, MA, USA: Athena Scientific, 2008. engineering.
He is currently a Professor with the Depart-
ment of Mechanical Engineering and also the Di-
Jaekwang Nam received the B.S. degree in me- rector of the Precision Rotating Electromechan-
chanical engineering from Hanyang University, ical Machine Laboratory, Hanyang University, Seoul, South Korea. His
Seoul, South Korea, in 2011, and he is currently current research is focused on microrobots actuated by magnetic nav-
working toward the Ph.D. degree in mechanical igation system, and electromechanical systems such as motors and
convergence engineering at Hanyang University, actuators. He has authored or coauthored more than 280 articles pub-
Seoul, South Korea. lished in journals and conferences in his field and more than 29 patents,
His research interest includes various struc- including several international patents.
tures of microrobots performing multifunctions in
human blood vessels and its magnetic naviga-
tion system.

Wonseo Lee (S’16) received the B.S. degree in


mechanical engineering in 2014 from Hanyang
University, Seoul, South Korea, where he is cur-
rently working toward the Ph.D. degree in me-
chanical convergence engineering.
His current research interest includes design,
analysis, and control of magnetic robots and
magnetic catheters with electromagnetic sys-
tems for biomedical applications.

You might also like