- Email: [email protected]
An electrorheological spherical joint actuator for a haptic master with application to robot-assisted cutting surgery
Sensors and Actuators A 249 (2016) 163–171
Contents lists available at ScienceDirect
Sensors and Actuators A: Physical journal homepage: www.elsevier.com/locate/sna
An electrorheological spherical joint actuator for a haptic master with application to robot-assisted cutting surgery Yong-Hoon Hwang, Seok-Rae Kang, Seung-Woo Cha, Seung-Bok Choi ∗ Smart Structures and Systems Laboratory, Department of Mechanical Engineering, Inha University, Incheon 402-751, Republic of Korea
a r t i c l e
i n f o
Article history: Received 6 April 2016 Received in revised form 30 August 2016 Accepted 30 August 2016 Available online 31 August 2016 Keywords: Electro-rheological (ER) fluid ER spherical joint actuator Haptic master Slave robot Cutting surgery Force feedback
a b s t r a c t This paper presents a spherical joint actuator using an electrorheological (ER) fluid to make a new haptic master which can be applicable to a robot-assisted cutting surgery. The proposed haptic master consists of two actuators for different motions; ER spherical joint device for 3-DOF rotational motion and ER linear device for 1-DOF translational motion. An appropriate size of the haptic master is designed based on the governing equations of the dynamic model and manufactured. The haptic master is then connected to a slave robot for the cutting surgery of a tissue. In the surgical process, the forces required to cut the tissue (a pork in this work) are obtained using a force sensor via experimets and used as the desired forces to be tracked by a slave robot controlled by a proportional-integrative-derivative (PID) controller. The desired repulsive forces are then embodied through the haptic master so that the operator can feel the forces occurred in cutting process (actual forces). In order to validate the effectiveness of the proposed haptic master, the tracking control performances between the desired force and actual force are evaluated in time domain. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Recently, a human-machine interface has become more and more important to achieve communication between humans and machine systems via human senses such as sight, hearing and touch. The interfaces with the third human sense, the sense of touch, are being actively researched in various application fields [1–3]. These systems which provide stimulus information such as kinesthetic force to a user are called haptic systems or simply haptics. Recently, a robot-assisted minimally invasive surgery has been introduced by incorporating with a haptic master device whose function is not only to generate the motion for a slave robot, but also to reflect some physical constraints to the haptic operator such as viscosity and stiffness of the touched tissues of bones and organs. A. Frisoli et al. [4] designed a force-based impedance control of a haptic master for a haptic interface system. A gripper with force feedback in minimally invasive surgery was also developed by M. MacFarlane et al. [5]. In addition, the haptic master for mobile manipulator has been proposed a lot [6,7]. In general, the haptic device requires an actuating component for the tactile feedback. Currently, a motor driven by an electronic circuit is a typical actuator for the haptic devices, and many physical haptic feedbacks are
∗ Corresponding author. E-mail address: [email protected] (S.-B. Choi). http://dx.doi.org/10.1016/j.sna.2016.08.033 0924-4247/© 2016 Elsevier B.V. All rights reserved.
carried in the form of vibration based on eccentric motors. The commercial eccentric motors generate vibration using an unbalanced mass. It has some disadvantages such as complex mechanism, difficulty in continuous and smooth force control and safety problems [8]. Therefore, the haptic feedback using the vibration motor limits the ability to discriminate the tactile force. To overcome this limitation, a voice coil motor (VCM)-type actuator has been introduced for the haptic devices; a linear resonance actuator using resonance frequency to generate vibrations [9]. However, this device has a limited application area in haptic feedback because the human sense of touch is far more sensitive than the senses of sight or hearing. A continuous and complex force feedback is still required in various robotic applications. Therefore, many researchers are working on a new actuating mechanism based on smart materials such as piezoelectric actuator, shape memory alloys, electro-active polymers, and so on [10]. Subsequently, several researches have been recently made by adopting smart fluids such as electro-rheological (ER) fluids and magneto-rheological (MR) fluids to overcome those obstacles. The yield stress of the smart fluids is easily changed by controlling the intensity of electric or magnetic field. Due to this phenomenological behavior, smart fluids have several benefits such as resistance to external forces or pressures, high stability owing to the viscous property and reliable control performance [11]. Li et al. [12] developed an MR fluid based haptic system featuring MR brake and gimbal structure. Senkal et al. [13] developed an MR spher-
164
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
ical brake and performed virtual test. Blake et al. [14] developed a force-feedback haptic glove which can allow the user to pick up and feel virtual objects. Kikuchi et al. [15] proposed a novel limb rehabilitation system using ER brake. Mavroidis et al. [16] proposed an ER haptic device called MEMICA (MEchanical MIrroring using Controlled stiffness and Actuators). Böse et al. [17] researched an ER fluid-based force-feedback joystick consisting of a ball and socket joint. The haptic feedback using smart fluids can give continuous, good stability and reliable force feedback. But, these previous studies on the haptic feedbacks using smart fluids are limited in their degree-of-freedom for dynamic motions. Subsequently, the haptic device capable of multi-degree-of-freedom and multi-motion force feedback is still required for effective applications such as surgical robots and rehabilitation devices. On the other hand, several researches on the application of ER or MR haptic master to surgical system have been undertaken. Ahmadkhanlou et al. developed haptic systems for teleoperational surgery using MR fluids for semi-active force feedback [18]. An & Kwon modeled MR actuators by considering magnetic hysteresis to determine the nonlinear torque-current relationship [19]. A haptic joystick based on MR brake mechanism has been also utilized by Bachman & Milecki [20], and Blake & Gurocak proposed a haptic glove that can obstruct finger movement [21]. A.M Okamura implemented several different types of tele-manipulation control laws, which can provide different capabilities for position, force, and environment impedance [22]. Reiley et al. achieved a successful visual force feedback resulted in reduced suture breakage, lower forces, and decreased force inconsistencies among novice robotic surgeons [23]. Scilingo et al. reported the possibility of using MR fluids to mimic compressional characteristics of biological tissues [24]. More recently, Kim et al. developed force modeling of various tissues and applied the model to robot-assisted cutting surgery using MR haptic master activated by bi-directional clutches. It is here noted that the bi-directional clutch based on the haptic master has a complex mechanism [25]. Therefore, a simpler actuator for the haptic master mechanism needs to be developed along with application to the robot-assisted surgery. Consequently, the main contribution of this work is to propose a simple actuator for the haptic device using an ER fluid and apply it to the robot-assisted surgery. In order to achieve this goal, 4-DOF haptic master consisting of a simple spherical joint actuator and a linear actuator utilizing ER fluid is devised and manufactured. For the design of the haptic master, the generated torque and force are mathematically analyzed based on the device geometry and Bingham characteristics of ER Fluid. The manufactured haptic master is then tested by investigating the generated force and torque as a function of an electric filed applied to ER actuators. Subsequently, the haptic master is connected to the slave robot to undertake a simple cutting surgery. The effectiveness of the haptic feedback system on the cutting surgery is verified by adopting a pork as a cutting tissue in which a part of tumor is intentionally indexed. The tumor is excised from normal tissue with and without the haptic force feedback action and the results are compared in time domain. 2. Design of ER actuators The most typical smart fluids are MR and ER fluids, which belong to a family of rheological materials that undergo rheological phase-change under the application of fields. These smart fluids, in general, have been considered as the Bingham fluid whose constitutive equation is given by [26,27]: = ˙ + y (·)
which is a function of field strength. The dynamic yield stress can be described by a simple exponential form or more complex form such as polynomials. In this work, a silica based ER fluid is used for 4-DOF haptic master system and its measured yield stress is determined in exponential form as follows [26]: y (E) = 484.463E 1.374
where E denotes the electric field with units of kV/mm. This equation will be used to calculate the generated force (or torque) from the haptic master (refer to equations (5) and (7)) Fig. 1(a) shows the configurations of a 4-DOF haptic force feedback device utilizing ER actuators. It is seen that two distinct actuating mechanisms are introduced to achieve both rotational and translational motions. One is the spherical joint which features a simple mechanism, but can generate 3 rotational motions and corresponding force reflections. The other is the linear actuating device which can generate translation force reflection. From the schematic configuration of the proposed ER spherical joint, it is seen that there exist the spherical ball and spherical housing as the inner and outer electrode, respectively. The ER fluid is fully filled between the inner and outer electrodes. The proposed piston type linear device is divided into the upper and lower chambers by the piston head. These chambers are fully filled with the ER fluid. When the piston moves, the ER fluid can be transferred from one chamber to the other. The inner and outer cylinders are playing as the positive and negative electrodes. Two columns are fixed between top and bottom plates. If the operator rotates the forceps in yawing direction, this motion is transmitted to bottom plate via fixed columns. The bottom plate is connected with operation link of ER spherical joint to transfer yawing motion of forceps to ER spherical joint. The most fascinating feature of the proposed actuating mechanism is that the force-feedback can be achieved along operator’s moving direction in a semi-active manner. The generated torque by the ER spherical joint actuator consists of controllable torque, TER , and viscous friction torque, T which can be expressed by surface integral as follows: TER = T =
r dA S y m S
(3)
r ˙ m dA
where S is the surface area of the spherical joint. rm is the distance of moment arm from the one point on the spherical surface to rotational axis. Fig. 2 shows a spherical coordinate system considering three rotational motions: pitching, rolling and yawing. An arbitrary point on the surface of electrode is described by the angles u andv. The moment arm for X, Y and Z axis are determined by xrm = re yrm = re
(sin v)2 + (cos v sin u)2
(4)
(sin v)2 + (cos v sin u)2
zrm = re cos v where re is the radius of the spherical joint. The generated controllable torque, viscous friction torque in three rotational motions are derived by surface integral as follows:
X
T ER
= (E)re − 2
3
0
Y
T ER
= (E)re 3 2 −
(1) Z
Where is the shear stress, is the dynamic viscosity, and is ˙ the shear rate. y (·) is the dynamic yield stress of the smart fluid,
(2)
T ER
=
v0
2
0
− v0 2 − v0 2
2
y (E)re 3
2
2
2
2
(sin v) + (cos v sin u) cos vdudv
0
2
y (E)re 3
(sin v) + (cos v sin u) cos vdudv
0
2
y (E)re 3 cos2 vdudv 0
(5)
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 1. The proposed 4-DOF Haptic Master.
165
166
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 2. Controllable torque and force with respect to the design.
X
Y
8ω 4 r − 3ts e
T =
8ω 4 T = r − 3ts e
Z
T =
v0
− 2
0
− v0 2
2
0
0
− v0 2
0
ω 4 r ts e
2
0
ω re4 ts
2
as summation of the elastic force of the gas chamber, the viscous friction force of ER fluid, and the controllable force of ER fluid as follows [28]:
2
(sin v) + (cos v sin u) cos2 vdudv
ZF ER 2
2
2
(sin v) + (cos v sin u) cos vdud
= P2 (Ap − Af ) − P1 (Ap − As − Af )
(6)
2
= Patm As + z Fn + z Fc sgn(x˙ p )
(7)
ω re4 cos3 vdud ts
In the above, ω is the angular velocity of the spherical joint and ts is the gap size. v0 is the constant angle of outer electrode. By assuming quasi-static behavior of the proposed device, the generated force of the proposed linear actuator device can be expressed
= Patm As +
6l Rtp3
(Ap − As − Af )2 x˙ p + (Ap − As − Af )
cl y sgn(x˙ p ) tp
Where, Ap , As and Af are the piston head area, the piston crosssectional area and column area, respectively. P1 and P2 are the
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
167
pressures in the upper and lower chambers of the proposed device, respectively. Patm is the pressure of atmosphere (1013.25 hPa). R is the average radius of the annular duct. c is the flow velocity profile coefficient which has a value ranging from 2.07 to 3.07. l is the stroke of proposed device, xp is the piston displacement, and tp is the gap size between the inner and outer cylinder. 3. Performance test of haptic master Among the generated torques in Eqs. (6) and (7), the controllable torque has the largest effect on the performance of the haptic master. It is desirable that the controllable torque takes a large value under geometric constraints to make a compact size. Therefore, the optimization problem is to maximize the controllable torque subject to two significant design parameters: re varied from 0 to 60 mm and v0 varied from 0 to 90◦ . For the design of two actuators, the significant design parameters concerned with the controllable force are stroke, gap size, and radius of the piston. In order to maximize the controllable force, the radii of the piston and column, which are in inversely proportion to the controllable force, are fixed by 1 mm. The gap size, tP, is set by1 mm. The maximum controllable force (applying 2 kV/mm electric field) is calculated by changing the design parameters: stroke, l, varied from 30 mm to 60 mm and radius of piston, rh , varied from 0 to 40 mm. Fig. 3 shows controllable force and torque of the proposed haptic master calculated on the basis of the principal design parameters such as electrode gap. The result shown in Fig. 3 (a) is obtained by the following conditions: design parameters are chosen to obtain proper workspace and desired magnitude of repulsive force level which is set as10N. So, l is used as 50 mm to achieve proper workspace. In order to obtain desired magnitude of the repulsive force level, the radius of the piston head, rh , is chosen by 15 mm. Figs. 3(b) and (c) show the torque of yawing motion and pitching (and rolling) motion, respectively. Outer electrode angle,v0 , is related with workspace of the spherical joint. In order to obtain proper workspace and compact size, v0 is determined by 45◦ . Then, in order to achieve the desired torque level of 0.6 Nm, re is determined as 42 mm. When angular velocity is assumed to be 1.5 rad/s, the viscous friction torque is calculated by 0.01Nm. This result means that controllable torque is a major portion of whole generated torques. Based on the analyzed results, the haptic master is manufactured as shown in Fig. 1(b). It is noted that the operation link is connected to the ER linear device. A 6-axis force/torque sensor (ATI, Nano 17) is installed in the middle of the operation link to measure generated force. Encoders (Autonics Corp., E40H) and linear variable differential transducer (LVDT, Ladian Corp., M150), are used to measure position information. The spherical measuring link is also connected to the operation link to achieve rotation information. It is remarked that the measured position information is converted to command signal to operate the slave robot. The position of the operational knob can be expressed in the spherical coordinate system as shown in Fig. 2(a). Then, the coordinates of the master system is derived from the coordinate image shown in Fig. 2(b) as follows. X = r(t) cos ˛x sin ˛y Y = r(t) sin ˛x
(8)
Z = r(t) cos ˛x cos ˛y where ˛x and ˛y are measured signals from encoders. r0 is the length between top of the inner electrode and top of the piston. l1 is the length between piston and LVDT and l2 (t) is the measured signal from LVDT. The length of operation link is calculated r(t) = r0 +
l2 (t)2 − l1
2
(9)
Fig. 3. Controllable torque and force with respect to the design parameters.
In order to evaluate the accuracy of the optimal design and torque models, the force reflection of the master is measured and compared with simulated one. The manufactured haptic master with force and position sensors is freely moved by a person (operator). The step responses of the haptic master are obtained for four motions (precession-nutation-spin-translation) and presented in Fig. 4. In this work, the step responses are obtained by measuring the time when the output signal is reached to 65% of the steady state value. From the results, the precession and nutation motions have quite similar responses, and their force levels are about 7.6 N and 6.5 N, respectively. The generated torque for the spin motion and force for the linear motion is about 0.31 Nm and 7.13 N, respectively. It is remarked here that the maximum achievable rotational angle along the pitching and rolling motions are 100◦ and the amount of ER fluid is 95 ml. Tables 1 and 2 show the dynamic properties and geometric specification of the proposed haptic master. It is noted that in order to achieve the results shown in Figs. 7 and 9, 20 times of experiments have been carried out for each direction and the average value is used to plot the results. 4. Kinematics of slave robot The slave robot adopted in this work is a vertically articulated type and has five DC servomotors (Dynamixel MX series,
168
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Table 2 Specifications of the ER haptic master. Specification
ER Haptic Master
Degrees of Freedom Maximum Achievable Rotational Angle Along Yawing motions Maximum Achievable Translation Displacement Maximum Achievable Rotational Angle Along Pitching and Rolling Motions Required Fluid Amount Actuating Mechanism Design Complexity Maximum Achievable Force/Torque Level
4 ∞ 50 mm 100◦ 95 ml Semi-active Simple 7.56 N, 0.29 Nm
Fig. 5. The slave robot for cutting surgery.
Fig. 4. Measured step responses of the 4-DOF ER haptic master.
Table 1 Dynamic properties of ER haptic master. Characteristics
Value
Damping Ratio of Precession Motion Natural Frequency of Precession Motion Damping Ratio of Nutation Motion Natural Frequency of Nutation Motion Damping Ratio of Spin Motion Natural Frequency of Spin Motion Time Constant of Translation Motion
0.603 29.38 rad/s 0.993 23.98 rad/s 0.948 8.63 rad/s 0.34 s
Robotis Corp.); four for positioning and one for spinning of the end-effector, directly installed inside the joints, as shown in Fig. 5. During surgery, the repulsive force from the tissue is measured by
Fig. 6. Coordinates of the slave robot.
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
169
a force sensor (Nano 17, ATI Industrial Automation, Inc.) attached at the tool tip. In fact, for real surgical environment the surgical instrument has to pass through a small fixed incision point on the patient’s abdomen during operation. Therefore, for safety, the surgical instrument goes through a ball joint on a fixed bar. This pivot point coincides with the incision point during surgery, and the 3◦ -of-freedom (DOF) rotational motions of the slave are realized through it. However, in this work the cutting surgery only which does not require small incision has been undertaken. A gimbal structure is attached to motor 5 to generate the spinning motion of the surgical instrument. Thus, this arrangement of the slave robot allows the surgical instrument to freely rotate in the roll, pitch, and spin directions with the roll and pitch motions pivoting about the ball-joint constraint of the fixed bar. The five servomotors are controlled by a proportional-integrative-derivative (PID) controllers, which do not require equations of motion of the system to be controlled. Therefore, the dynamic analysis of the slave robot is not required to manipulate its position; the kinematic analysis only is sufficient. Assuming the required position of the slave’s tool tip as (x, y, z) corresponding to the master position (X, Y, Z) in Eq. (8), the coordinates of the point P with respect to the reference frame shown in Fig. 6 can be obtained as follows: x = [l1 cos 2 + l2 cos(2 − 3 ) + l3 ] cos 1 y = [l1 cos 2 + l2 cos(2 − 3 ) + l3 ] sin 1
(10)
z = l1 sin 2 + l2 sin(2 − 3 ) Subsequently, from Eq. (10), the angles (1, 2, 3 ) of the slave robot to achieve the desired motion of the end-effector are calculated through the inverse kinematics analysis as follows: 1 = a tan 2
y x
,
2 = a tan 2
3 = a tan 2 4 =
2
c
− a tan 2
a2 + b2 − c 2
a
x cos 1 + y sin 1 − l2 cos 2 l1 + l2 cos 2 − dz
b
,
− 2 ,
Fig. 7. Position control and error of the slave robot.
(11) − 2 − 3 ,
jectory using the microprocessor. The actual each positions of the tool tip accurately followed the desired position trajectory is shown in Fig. 7(a). It is identified that the tracking errors of each directional position are less than 0.005 mm. Therefore, x position error only is presented as shown in Fig. 7(b) because the tracking errors of the other directions show similar tendency to x direction in terms of the error magnitude and the profile.
where a = −2l2 (x cos 1 + y sin 1 ), b = 2l2 (l1 − z),
2
c = l32 − (x cos 1 + y sin 1 ) + l12 − 2l1 dz + l22 − z 2
And, the end point of the slave robot, E(x’, y’, z’) can be determined as follows: x =
l4 l5 − (l4 − OP)[l1 cos 2 + l2 cos(2 − 3 ) + l3 ] cos 1 OP
y = −
(l4 − OP)[l1 cos 2 + l2 cos(2 − 3 ) + l3 ] sin 1
(12)
OP z = −
(l4 − OP)[l1 sin 2 + l2 sin(2 − 3 )] OP
In the above, l1 , l2 , and l3 are the lengths of links 1, 2, and 3, respectively, and link 1 is connected to the base. 1 is the rotational angle between the arm and the body, 2 is the angle between link 1 and the xy-plane, 3 is the supplement of the angle between link 1 and link 2, and 4 is the supplement of the angle between link 2 and link 3. The precision of the slave’s motion, according to the kinematic analysis is tested by randomly generating a desired tra-
5. Cutting surgery The purpose of the experiment is to evaluate the performance in a practical surgery task owing to the augmentation of the haptic feedback control. A tumor-cutting task is given to two male participants (operators) who repeat each experiment 20 times. The performance is then assessed by how precisely the operator can cut the tumor using the surgery system shown in Fig. 8. As seen from the figure when the operator manipulates the haptic master, the motion commands are transferred to the slave robot via dSPACE (analog/digital-digital/analog converter, PID controller: NI PXIe-6363). The movement of the master is measured by encoders (E40H8-1000-3-N-5, Autonics Corp.), and its frictional torques are determined by torque sensors (SDS-100K, Sensetech Corp.). While performing the surgical tasks according to the motion commands, the slave robot transmits the force data from the force sensor (Nano 17, ATI Corp.), which is installed in the middle of the instrument. The force sensor measures the contact force that the slave robot
170
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 8. Control block diagram for cutting surgery.
perceives, and the haptic master allows the operator to feel this measured force through the degree of solidification of ER fluid. In the feedback process, the operator can identify the difference between actual force occurred during the cutting surgery and the desired force stored in the computer (microprocessor). Prior to performing the surgical task, the proposed system is tested by comparing the desired force and actual force which reflects the generated force from the haptic master. This test can provide the information how the operator precisely manipulates the slave robot based on the haptic sensation feedback signal (repulsive force). In addition, this information is significant how
accurately the tumor tissue is excised from the normal tissue. Fig. 9 shows the generated forces from the haptic master during the cutting surgery in the x-, y-, z-, and yaw direction, respectively. It is identified that the average force error does not exceed 0.038 N, and the standard deviation is 0.022 N. It is seen from Fig. 9(c) that the error in z direction is suddenly increased during the transient motion changing the direction from downward to upward. This error is caused by the gravity effect which is not considered in the dynamic model of the slave robot. It is noted that this error can be reduced by adopting gravity compensator integrated with the controller. Now, the cutting surgery is undertaken by adopting the tissue as a piece of pork in which the intact tissue is partially dyed in black color to mark the tumor part on it. Fig. 10 shows the photograph of the original and cut tissues achieved under two different conditions: without the haptic feedback and with the haptic feedback control action. It is noted the surgery without the haptic feedback is undertaken by directly observing the cutting tissue without any sensation information from the haptic master. It is clearly observed that more accurate cutting surgery is obtained with the haptic feedback control as expected. The preliminary surgical results presented in this work are quite self-explanatory justifying that the proposed method can possibly apply to real robot-assisted minimally invasive surgery (MIS) after augmenting more sophisticated control logics and teleoperated function. 6. Conclusions In this work, a new type of the haptic master which is applicable to the robot-assisted cutting surgery was devised using ER actuators and its effectiveness was validated by experimental tests. The proposed haptic master and slave robot were designed and manufactured using the dynamic models of ER spherical and linear actuators and kinematic model of the slave robot. Prior to performing the cutting surgery, the effectiveness of the proposed
Fig. 9. Comparison of desired force and actual force occurred from the cutting surgery.
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 10. Tumor removal with and without feedback control.
haptic master was proved showing excellent agreement between the desired and actual forces in all manipulation directions. For example, it has been identified that the average force error does not exceed 0.038 N, and the standard deviation is 0.022 N. This accurate results provide the information how the operator precisely manipulates the slave robot based on the haptic sensation feedback signal (repulsive force). The proposed haptic master was connected to the slave robot in order to realize a simple surgery: cutting a pork. In order to observe the difference between with and without the haptic feedback signal, the part of the tissue (pork) was indexed by black ink. It has been demonstrated by experiment that more accurate cutting surgery can be achieved with the haptic feedback signal as expected. This indicates that the proposed haptic master featuring simple ER actuators can be one of potential candidates for the robot-assisted MIS in medical applications. It is noted here that the proposed ER haptic master doesn’t have any motion limitation. In addition, the durability of accuracy depends on the characteristics of ER fluid. It has been proved that there is no change of material characteristics of ER fluid under temperature of 100 ◦ C and the durability is guaranteed up to 106 cycles. However, a caution during the operation is required since a high voltage (about 2 kV) needs to be applied to get required force and torque during the surgery. Normally, the input current (about 2 mA) is very low, and hence the power to get the required force or torque is less than 5W for the surgical operation. It should be also remarked here that this work doesn’t contain how to control force feedback for the teleoperation which is crucial issue for final commercialization. The study on the distance-based teleoperation for the force feedback control of the proposed haptic master will be carried out in the near future as a second phase of this work. Acknowledgement This work was supported by National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (No. 2015R1A2A1A5054000). This financial support is gratefully acknowledged. References [1] Intuitive surgical, http://www.intuitivesurgical.com. [2] V. Hayward, O.R. Astley, M. Cruz-Hernandez, D. Grant, G. Robles-De-La-Torre, Haptic interfaces and devices, Sens. Rev. 24 (2004) 16–29. [3] J. Aoki, T. Murakami, A method of road condition estimation and feedback utilizing haptic pedal Advanced Motion Control, 2008. AMC’08. 10th IEEE International Workshop on, IEEE, 2008 pp. 777–782.
171
[4] A. Frisoli, E. Sotgiu, C. Avizzano, D. Checcacci, M. Bergamasco, Force-based impedance control of a haptic master system for teleoperation, Sens. Rev. 24 (2004) 42–50. [5] M. MacFarlane, J. Rosen, B. Hannaford, C. Pellegrini, M. Sinanan, Force-feedback grasper helps restore sense of touch in minimally invasive surgery, J. Gastrointest. Surg. 3 (1999) 278–285. [6] D. Ryu, C. Cho, M. Kim, J.-B. Song, Design of a 6 DOF haptic master for teleoperation of a mobile manipulator, Robotics and Automation, 2003. Proceedings. ICRA’03. IEEE International Conference on, IEEE, (2003) 3243–3248. [7] N. Diolaiti, C. Melchiorri, Teleoperation of a mobile robot through haptic feedback, Haptic Virtual Environments and Their Applications, IEEE International Workshop 2002 HAVE, IEEE (2002) 67–72. [8] F. Pierrot, E. Dombre, E. Dégoulange, L. Urbain, P. Caron, S. Boudet, J. Gariépy, J.-L. Mégnien, Hippocrate: a safe robot arm for medical applications with force feedback, Med. Image Anal. 3 (1999) 285–300. [9] S. Do Kweon, I.O. Park, Y.H. Son, J. Choi, H.Y. Oh, Linear vibration motor using resonance frequency, Google Patents, 2008. [10] A. Baz, K. Imam, J. McCoy, Active vibration control of flexible beams using shape memory actuators, J. Sound Vib. 140 (1990) 437–456. [11] S.-S. Han, S.-B. Choi, C.-C. Cheong, Position control of X–Y table mechanism using electro-rheological clutches, Mech. Mach. Theory 35 (2000) 1563–1577. [12] W. Li, B. Liu, P.B. Kosasih, X. Zhang, A 2-DOF MR actuator joystick for virtual reality applications, Sens. Actuators A: Phys. 137 (2007) 308–320. [13] D. Senkal, H. Gurocak, Spherical brake with MR fluid as multi degree of freedom actuator for haptics, J. Intell. Mater. Syst. Struct. 20 (2009) 2149–2160. [14] J. Blake, H.B. Gurocak, Haptic glove with MR brakes for virtual reality, Mechatron. IEEE/ASME Trans. 14 (2009) 606–615. [15] T. Kikuchi, K. Fukushima, J. Furusho, T. Ozawa, Development of Quasi-3DOF upper limb rehabilitation system using ER brake: PLEMO-P1, J. Phys.: Conf. Ser. (2009) 012015, IOP Publishing. [16] C. Mavroidis, C. Pfeiffer, J. Celestino, Y. Bar-Cohen, Design and modeling of an electro-rheological fluid based haptic interface, Proceedings of the 2000 ASME Mechanisms and Robotics Conference (2000) 10–13. [17] H. Böse, M. Baumann, G. Monkman, S. Egersdörfer, A. Tunayar, H. Freimuth, H. Ermert, W. Khaled, A new ER fluid based haptic actuator system for virtual reality, Int. J. Mod. Phys. B 19 (2005) 1628–1634. [18] F. Ahmadkhanlou, G. Washington, S.E. Bechtel, Modeling and control of single and two degree of freedom magnetorheological fluid-Based haptic systems for telerobotic surgery, J. Intell. Mater. Syst. Struct. (2009). [19] J. An, D.-S. Kwon, Modeling of a magnetorheological actuator including magnetic hysteresis, J. Intell. Mater. Syst. Struct. 14 (9) (2003) 541–550. [20] P. Bachman, A. Milecki, MR haptic joystick in control of virtual servo drive, J. Phys.: Conf. Ser. 149 (1) (2009) 012034. [21] J. Blake, H.B. Gurocak, Haptic glove with MR brakes for virtual reality, Mechatron. IEEE/ASME Trans. 14 (5) (2009) 606–615. [22] A.M. Okamura, Methods for haptic feedback in teleoperated robot-assisted surgery, Ind. Robot: Int. J. 31 (6) (2004) 499–508. [23] Reiley, Takintope Akinbiyi, Darius Burschka, David C. Chang, Allison M. Okamura, David D. Yuh, Effects of visual force feedback on robot-assisted surgical task performance, J. Thorac. Cardiovasc. Surg. 135 (2008) 196–202. [24] E.P. Scilingo, A. Bicchi, D. De-Rossi, A. Scotto, A magnetorheological fluid as a haptic display to replicate perceived biological tissues compliance, Microtechnologies in Medicine and Biology, 1st Annual International, Conference On 2000 (2000) 229–233. [25] P.H. Kim, S.M. Kim, Y.D. Park, S.B. Choi, Force modeling for incersions into various tissues with MRF haptic master, Smart Mater. Struct. 25 (3) (2016) 13, Article No. 035008. [26] S. Choi, D. Lee, Rotational motion control of a washing machine using electrorheological clutches and brakes, Proc. Inst. Mech. Eng. Part C: J. Mech. Eng. Sci. 219 (2005) 627–637. [27] S.G. Lim, J.S. Oh, P.B. Nguyen, S.B. Choi, A medical haptic master featuring bi-directional clutch and brake using electrorheological fluid, Adv. Sci. Lett. 13 (1) (2012) 322–326. [28] Q.-H. Nguyen, S.-B. Choi, A new approach for dynamic modeling of an electrorheological damper using a lumped parameter method, Smart Mater. Struct. 18 (2009) 115020.
Biography
Seung-Bok Choi is a Professor of Inha University and Director of Smart Structures and Systems Laboratory. He received his B.S. degree in Mechanical engineering from Inha University, Korea in 1979, and his M.S. and Ph.D. degrees in Mechanical engineering from Michigan State University, USA in 1986 and 1990 respectively. His research is focused on the applications of magnetorheolgocial, electrorheological fluids, piezoelectric materials and shape memory alloys.
Contents lists available at ScienceDirect
Sensors and Actuators A: Physical journal homepage: www.elsevier.com/locate/sna
An electrorheological spherical joint actuator for a haptic master with application to robot-assisted cutting surgery Yong-Hoon Hwang, Seok-Rae Kang, Seung-Woo Cha, Seung-Bok Choi ∗ Smart Structures and Systems Laboratory, Department of Mechanical Engineering, Inha University, Incheon 402-751, Republic of Korea
a r t i c l e
i n f o
Article history: Received 6 April 2016 Received in revised form 30 August 2016 Accepted 30 August 2016 Available online 31 August 2016 Keywords: Electro-rheological (ER) fluid ER spherical joint actuator Haptic master Slave robot Cutting surgery Force feedback
a b s t r a c t This paper presents a spherical joint actuator using an electrorheological (ER) fluid to make a new haptic master which can be applicable to a robot-assisted cutting surgery. The proposed haptic master consists of two actuators for different motions; ER spherical joint device for 3-DOF rotational motion and ER linear device for 1-DOF translational motion. An appropriate size of the haptic master is designed based on the governing equations of the dynamic model and manufactured. The haptic master is then connected to a slave robot for the cutting surgery of a tissue. In the surgical process, the forces required to cut the tissue (a pork in this work) are obtained using a force sensor via experimets and used as the desired forces to be tracked by a slave robot controlled by a proportional-integrative-derivative (PID) controller. The desired repulsive forces are then embodied through the haptic master so that the operator can feel the forces occurred in cutting process (actual forces). In order to validate the effectiveness of the proposed haptic master, the tracking control performances between the desired force and actual force are evaluated in time domain. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Recently, a human-machine interface has become more and more important to achieve communication between humans and machine systems via human senses such as sight, hearing and touch. The interfaces with the third human sense, the sense of touch, are being actively researched in various application fields [1–3]. These systems which provide stimulus information such as kinesthetic force to a user are called haptic systems or simply haptics. Recently, a robot-assisted minimally invasive surgery has been introduced by incorporating with a haptic master device whose function is not only to generate the motion for a slave robot, but also to reflect some physical constraints to the haptic operator such as viscosity and stiffness of the touched tissues of bones and organs. A. Frisoli et al. [4] designed a force-based impedance control of a haptic master for a haptic interface system. A gripper with force feedback in minimally invasive surgery was also developed by M. MacFarlane et al. [5]. In addition, the haptic master for mobile manipulator has been proposed a lot [6,7]. In general, the haptic device requires an actuating component for the tactile feedback. Currently, a motor driven by an electronic circuit is a typical actuator for the haptic devices, and many physical haptic feedbacks are
∗ Corresponding author. E-mail address: [email protected] (S.-B. Choi). http://dx.doi.org/10.1016/j.sna.2016.08.033 0924-4247/© 2016 Elsevier B.V. All rights reserved.
carried in the form of vibration based on eccentric motors. The commercial eccentric motors generate vibration using an unbalanced mass. It has some disadvantages such as complex mechanism, difficulty in continuous and smooth force control and safety problems [8]. Therefore, the haptic feedback using the vibration motor limits the ability to discriminate the tactile force. To overcome this limitation, a voice coil motor (VCM)-type actuator has been introduced for the haptic devices; a linear resonance actuator using resonance frequency to generate vibrations [9]. However, this device has a limited application area in haptic feedback because the human sense of touch is far more sensitive than the senses of sight or hearing. A continuous and complex force feedback is still required in various robotic applications. Therefore, many researchers are working on a new actuating mechanism based on smart materials such as piezoelectric actuator, shape memory alloys, electro-active polymers, and so on [10]. Subsequently, several researches have been recently made by adopting smart fluids such as electro-rheological (ER) fluids and magneto-rheological (MR) fluids to overcome those obstacles. The yield stress of the smart fluids is easily changed by controlling the intensity of electric or magnetic field. Due to this phenomenological behavior, smart fluids have several benefits such as resistance to external forces or pressures, high stability owing to the viscous property and reliable control performance [11]. Li et al. [12] developed an MR fluid based haptic system featuring MR brake and gimbal structure. Senkal et al. [13] developed an MR spher-
164
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
ical brake and performed virtual test. Blake et al. [14] developed a force-feedback haptic glove which can allow the user to pick up and feel virtual objects. Kikuchi et al. [15] proposed a novel limb rehabilitation system using ER brake. Mavroidis et al. [16] proposed an ER haptic device called MEMICA (MEchanical MIrroring using Controlled stiffness and Actuators). Böse et al. [17] researched an ER fluid-based force-feedback joystick consisting of a ball and socket joint. The haptic feedback using smart fluids can give continuous, good stability and reliable force feedback. But, these previous studies on the haptic feedbacks using smart fluids are limited in their degree-of-freedom for dynamic motions. Subsequently, the haptic device capable of multi-degree-of-freedom and multi-motion force feedback is still required for effective applications such as surgical robots and rehabilitation devices. On the other hand, several researches on the application of ER or MR haptic master to surgical system have been undertaken. Ahmadkhanlou et al. developed haptic systems for teleoperational surgery using MR fluids for semi-active force feedback [18]. An & Kwon modeled MR actuators by considering magnetic hysteresis to determine the nonlinear torque-current relationship [19]. A haptic joystick based on MR brake mechanism has been also utilized by Bachman & Milecki [20], and Blake & Gurocak proposed a haptic glove that can obstruct finger movement [21]. A.M Okamura implemented several different types of tele-manipulation control laws, which can provide different capabilities for position, force, and environment impedance [22]. Reiley et al. achieved a successful visual force feedback resulted in reduced suture breakage, lower forces, and decreased force inconsistencies among novice robotic surgeons [23]. Scilingo et al. reported the possibility of using MR fluids to mimic compressional characteristics of biological tissues [24]. More recently, Kim et al. developed force modeling of various tissues and applied the model to robot-assisted cutting surgery using MR haptic master activated by bi-directional clutches. It is here noted that the bi-directional clutch based on the haptic master has a complex mechanism [25]. Therefore, a simpler actuator for the haptic master mechanism needs to be developed along with application to the robot-assisted surgery. Consequently, the main contribution of this work is to propose a simple actuator for the haptic device using an ER fluid and apply it to the robot-assisted surgery. In order to achieve this goal, 4-DOF haptic master consisting of a simple spherical joint actuator and a linear actuator utilizing ER fluid is devised and manufactured. For the design of the haptic master, the generated torque and force are mathematically analyzed based on the device geometry and Bingham characteristics of ER Fluid. The manufactured haptic master is then tested by investigating the generated force and torque as a function of an electric filed applied to ER actuators. Subsequently, the haptic master is connected to the slave robot to undertake a simple cutting surgery. The effectiveness of the haptic feedback system on the cutting surgery is verified by adopting a pork as a cutting tissue in which a part of tumor is intentionally indexed. The tumor is excised from normal tissue with and without the haptic force feedback action and the results are compared in time domain. 2. Design of ER actuators The most typical smart fluids are MR and ER fluids, which belong to a family of rheological materials that undergo rheological phase-change under the application of fields. These smart fluids, in general, have been considered as the Bingham fluid whose constitutive equation is given by [26,27]: = ˙ + y (·)
which is a function of field strength. The dynamic yield stress can be described by a simple exponential form or more complex form such as polynomials. In this work, a silica based ER fluid is used for 4-DOF haptic master system and its measured yield stress is determined in exponential form as follows [26]: y (E) = 484.463E 1.374
where E denotes the electric field with units of kV/mm. This equation will be used to calculate the generated force (or torque) from the haptic master (refer to equations (5) and (7)) Fig. 1(a) shows the configurations of a 4-DOF haptic force feedback device utilizing ER actuators. It is seen that two distinct actuating mechanisms are introduced to achieve both rotational and translational motions. One is the spherical joint which features a simple mechanism, but can generate 3 rotational motions and corresponding force reflections. The other is the linear actuating device which can generate translation force reflection. From the schematic configuration of the proposed ER spherical joint, it is seen that there exist the spherical ball and spherical housing as the inner and outer electrode, respectively. The ER fluid is fully filled between the inner and outer electrodes. The proposed piston type linear device is divided into the upper and lower chambers by the piston head. These chambers are fully filled with the ER fluid. When the piston moves, the ER fluid can be transferred from one chamber to the other. The inner and outer cylinders are playing as the positive and negative electrodes. Two columns are fixed between top and bottom plates. If the operator rotates the forceps in yawing direction, this motion is transmitted to bottom plate via fixed columns. The bottom plate is connected with operation link of ER spherical joint to transfer yawing motion of forceps to ER spherical joint. The most fascinating feature of the proposed actuating mechanism is that the force-feedback can be achieved along operator’s moving direction in a semi-active manner. The generated torque by the ER spherical joint actuator consists of controllable torque, TER , and viscous friction torque, T which can be expressed by surface integral as follows: TER = T =
r dA S y m S
(3)
r ˙ m dA
where S is the surface area of the spherical joint. rm is the distance of moment arm from the one point on the spherical surface to rotational axis. Fig. 2 shows a spherical coordinate system considering three rotational motions: pitching, rolling and yawing. An arbitrary point on the surface of electrode is described by the angles u andv. The moment arm for X, Y and Z axis are determined by xrm = re yrm = re
(sin v)2 + (cos v sin u)2
(4)
(sin v)2 + (cos v sin u)2
zrm = re cos v where re is the radius of the spherical joint. The generated controllable torque, viscous friction torque in three rotational motions are derived by surface integral as follows:
X
T ER
= (E)re − 2
3
0
Y
T ER
= (E)re 3 2 −
(1) Z
Where is the shear stress, is the dynamic viscosity, and is ˙ the shear rate. y (·) is the dynamic yield stress of the smart fluid,
(2)
T ER
=
v0
2
0
− v0 2 − v0 2
2
y (E)re 3
2
2
2
2
(sin v) + (cos v sin u) cos vdudv
0
2
y (E)re 3
(sin v) + (cos v sin u) cos vdudv
0
2
y (E)re 3 cos2 vdudv 0
(5)
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 1. The proposed 4-DOF Haptic Master.
165
166
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 2. Controllable torque and force with respect to the design.
X
Y
8ω 4 r − 3ts e
T =
8ω 4 T = r − 3ts e
Z
T =
v0
− 2
0
− v0 2
2
0
0
− v0 2
0
ω 4 r ts e
2
0
ω re4 ts
2
as summation of the elastic force of the gas chamber, the viscous friction force of ER fluid, and the controllable force of ER fluid as follows [28]:
2
(sin v) + (cos v sin u) cos2 vdudv
ZF ER 2
2
2
(sin v) + (cos v sin u) cos vdud
= P2 (Ap − Af ) − P1 (Ap − As − Af )
(6)
2
= Patm As + z Fn + z Fc sgn(x˙ p )
(7)
ω re4 cos3 vdud ts
In the above, ω is the angular velocity of the spherical joint and ts is the gap size. v0 is the constant angle of outer electrode. By assuming quasi-static behavior of the proposed device, the generated force of the proposed linear actuator device can be expressed
= Patm As +
6l Rtp3
(Ap − As − Af )2 x˙ p + (Ap − As − Af )
cl y sgn(x˙ p ) tp
Where, Ap , As and Af are the piston head area, the piston crosssectional area and column area, respectively. P1 and P2 are the
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
167
pressures in the upper and lower chambers of the proposed device, respectively. Patm is the pressure of atmosphere (1013.25 hPa). R is the average radius of the annular duct. c is the flow velocity profile coefficient which has a value ranging from 2.07 to 3.07. l is the stroke of proposed device, xp is the piston displacement, and tp is the gap size between the inner and outer cylinder. 3. Performance test of haptic master Among the generated torques in Eqs. (6) and (7), the controllable torque has the largest effect on the performance of the haptic master. It is desirable that the controllable torque takes a large value under geometric constraints to make a compact size. Therefore, the optimization problem is to maximize the controllable torque subject to two significant design parameters: re varied from 0 to 60 mm and v0 varied from 0 to 90◦ . For the design of two actuators, the significant design parameters concerned with the controllable force are stroke, gap size, and radius of the piston. In order to maximize the controllable force, the radii of the piston and column, which are in inversely proportion to the controllable force, are fixed by 1 mm. The gap size, tP, is set by1 mm. The maximum controllable force (applying 2 kV/mm electric field) is calculated by changing the design parameters: stroke, l, varied from 30 mm to 60 mm and radius of piston, rh , varied from 0 to 40 mm. Fig. 3 shows controllable force and torque of the proposed haptic master calculated on the basis of the principal design parameters such as electrode gap. The result shown in Fig. 3 (a) is obtained by the following conditions: design parameters are chosen to obtain proper workspace and desired magnitude of repulsive force level which is set as10N. So, l is used as 50 mm to achieve proper workspace. In order to obtain desired magnitude of the repulsive force level, the radius of the piston head, rh , is chosen by 15 mm. Figs. 3(b) and (c) show the torque of yawing motion and pitching (and rolling) motion, respectively. Outer electrode angle,v0 , is related with workspace of the spherical joint. In order to obtain proper workspace and compact size, v0 is determined by 45◦ . Then, in order to achieve the desired torque level of 0.6 Nm, re is determined as 42 mm. When angular velocity is assumed to be 1.5 rad/s, the viscous friction torque is calculated by 0.01Nm. This result means that controllable torque is a major portion of whole generated torques. Based on the analyzed results, the haptic master is manufactured as shown in Fig. 1(b). It is noted that the operation link is connected to the ER linear device. A 6-axis force/torque sensor (ATI, Nano 17) is installed in the middle of the operation link to measure generated force. Encoders (Autonics Corp., E40H) and linear variable differential transducer (LVDT, Ladian Corp., M150), are used to measure position information. The spherical measuring link is also connected to the operation link to achieve rotation information. It is remarked that the measured position information is converted to command signal to operate the slave robot. The position of the operational knob can be expressed in the spherical coordinate system as shown in Fig. 2(a). Then, the coordinates of the master system is derived from the coordinate image shown in Fig. 2(b) as follows. X = r(t) cos ˛x sin ˛y Y = r(t) sin ˛x
(8)
Z = r(t) cos ˛x cos ˛y where ˛x and ˛y are measured signals from encoders. r0 is the length between top of the inner electrode and top of the piston. l1 is the length between piston and LVDT and l2 (t) is the measured signal from LVDT. The length of operation link is calculated r(t) = r0 +
l2 (t)2 − l1
2
(9)
Fig. 3. Controllable torque and force with respect to the design parameters.
In order to evaluate the accuracy of the optimal design and torque models, the force reflection of the master is measured and compared with simulated one. The manufactured haptic master with force and position sensors is freely moved by a person (operator). The step responses of the haptic master are obtained for four motions (precession-nutation-spin-translation) and presented in Fig. 4. In this work, the step responses are obtained by measuring the time when the output signal is reached to 65% of the steady state value. From the results, the precession and nutation motions have quite similar responses, and their force levels are about 7.6 N and 6.5 N, respectively. The generated torque for the spin motion and force for the linear motion is about 0.31 Nm and 7.13 N, respectively. It is remarked here that the maximum achievable rotational angle along the pitching and rolling motions are 100◦ and the amount of ER fluid is 95 ml. Tables 1 and 2 show the dynamic properties and geometric specification of the proposed haptic master. It is noted that in order to achieve the results shown in Figs. 7 and 9, 20 times of experiments have been carried out for each direction and the average value is used to plot the results. 4. Kinematics of slave robot The slave robot adopted in this work is a vertically articulated type and has five DC servomotors (Dynamixel MX series,
168
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Table 2 Specifications of the ER haptic master. Specification
ER Haptic Master
Degrees of Freedom Maximum Achievable Rotational Angle Along Yawing motions Maximum Achievable Translation Displacement Maximum Achievable Rotational Angle Along Pitching and Rolling Motions Required Fluid Amount Actuating Mechanism Design Complexity Maximum Achievable Force/Torque Level
4 ∞ 50 mm 100◦ 95 ml Semi-active Simple 7.56 N, 0.29 Nm
Fig. 5. The slave robot for cutting surgery.
Fig. 4. Measured step responses of the 4-DOF ER haptic master.
Table 1 Dynamic properties of ER haptic master. Characteristics
Value
Damping Ratio of Precession Motion Natural Frequency of Precession Motion Damping Ratio of Nutation Motion Natural Frequency of Nutation Motion Damping Ratio of Spin Motion Natural Frequency of Spin Motion Time Constant of Translation Motion
0.603 29.38 rad/s 0.993 23.98 rad/s 0.948 8.63 rad/s 0.34 s
Robotis Corp.); four for positioning and one for spinning of the end-effector, directly installed inside the joints, as shown in Fig. 5. During surgery, the repulsive force from the tissue is measured by
Fig. 6. Coordinates of the slave robot.
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
169
a force sensor (Nano 17, ATI Industrial Automation, Inc.) attached at the tool tip. In fact, for real surgical environment the surgical instrument has to pass through a small fixed incision point on the patient’s abdomen during operation. Therefore, for safety, the surgical instrument goes through a ball joint on a fixed bar. This pivot point coincides with the incision point during surgery, and the 3◦ -of-freedom (DOF) rotational motions of the slave are realized through it. However, in this work the cutting surgery only which does not require small incision has been undertaken. A gimbal structure is attached to motor 5 to generate the spinning motion of the surgical instrument. Thus, this arrangement of the slave robot allows the surgical instrument to freely rotate in the roll, pitch, and spin directions with the roll and pitch motions pivoting about the ball-joint constraint of the fixed bar. The five servomotors are controlled by a proportional-integrative-derivative (PID) controllers, which do not require equations of motion of the system to be controlled. Therefore, the dynamic analysis of the slave robot is not required to manipulate its position; the kinematic analysis only is sufficient. Assuming the required position of the slave’s tool tip as (x, y, z) corresponding to the master position (X, Y, Z) in Eq. (8), the coordinates of the point P with respect to the reference frame shown in Fig. 6 can be obtained as follows: x = [l1 cos 2 + l2 cos(2 − 3 ) + l3 ] cos 1 y = [l1 cos 2 + l2 cos(2 − 3 ) + l3 ] sin 1
(10)
z = l1 sin 2 + l2 sin(2 − 3 ) Subsequently, from Eq. (10), the angles (1, 2, 3 ) of the slave robot to achieve the desired motion of the end-effector are calculated through the inverse kinematics analysis as follows: 1 = a tan 2
y x
,
2 = a tan 2
3 = a tan 2 4 =
2
c
− a tan 2
a2 + b2 − c 2
a
x cos 1 + y sin 1 − l2 cos 2 l1 + l2 cos 2 − dz
b
,
− 2 ,
Fig. 7. Position control and error of the slave robot.
(11) − 2 − 3 ,
jectory using the microprocessor. The actual each positions of the tool tip accurately followed the desired position trajectory is shown in Fig. 7(a). It is identified that the tracking errors of each directional position are less than 0.005 mm. Therefore, x position error only is presented as shown in Fig. 7(b) because the tracking errors of the other directions show similar tendency to x direction in terms of the error magnitude and the profile.
where a = −2l2 (x cos 1 + y sin 1 ), b = 2l2 (l1 − z),
2
c = l32 − (x cos 1 + y sin 1 ) + l12 − 2l1 dz + l22 − z 2
And, the end point of the slave robot, E(x’, y’, z’) can be determined as follows: x =
l4 l5 − (l4 − OP)[l1 cos 2 + l2 cos(2 − 3 ) + l3 ] cos 1 OP
y = −
(l4 − OP)[l1 cos 2 + l2 cos(2 − 3 ) + l3 ] sin 1
(12)
OP z = −
(l4 − OP)[l1 sin 2 + l2 sin(2 − 3 )] OP
In the above, l1 , l2 , and l3 are the lengths of links 1, 2, and 3, respectively, and link 1 is connected to the base. 1 is the rotational angle between the arm and the body, 2 is the angle between link 1 and the xy-plane, 3 is the supplement of the angle between link 1 and link 2, and 4 is the supplement of the angle between link 2 and link 3. The precision of the slave’s motion, according to the kinematic analysis is tested by randomly generating a desired tra-
5. Cutting surgery The purpose of the experiment is to evaluate the performance in a practical surgery task owing to the augmentation of the haptic feedback control. A tumor-cutting task is given to two male participants (operators) who repeat each experiment 20 times. The performance is then assessed by how precisely the operator can cut the tumor using the surgery system shown in Fig. 8. As seen from the figure when the operator manipulates the haptic master, the motion commands are transferred to the slave robot via dSPACE (analog/digital-digital/analog converter, PID controller: NI PXIe-6363). The movement of the master is measured by encoders (E40H8-1000-3-N-5, Autonics Corp.), and its frictional torques are determined by torque sensors (SDS-100K, Sensetech Corp.). While performing the surgical tasks according to the motion commands, the slave robot transmits the force data from the force sensor (Nano 17, ATI Corp.), which is installed in the middle of the instrument. The force sensor measures the contact force that the slave robot
170
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 8. Control block diagram for cutting surgery.
perceives, and the haptic master allows the operator to feel this measured force through the degree of solidification of ER fluid. In the feedback process, the operator can identify the difference between actual force occurred during the cutting surgery and the desired force stored in the computer (microprocessor). Prior to performing the surgical task, the proposed system is tested by comparing the desired force and actual force which reflects the generated force from the haptic master. This test can provide the information how the operator precisely manipulates the slave robot based on the haptic sensation feedback signal (repulsive force). In addition, this information is significant how
accurately the tumor tissue is excised from the normal tissue. Fig. 9 shows the generated forces from the haptic master during the cutting surgery in the x-, y-, z-, and yaw direction, respectively. It is identified that the average force error does not exceed 0.038 N, and the standard deviation is 0.022 N. It is seen from Fig. 9(c) that the error in z direction is suddenly increased during the transient motion changing the direction from downward to upward. This error is caused by the gravity effect which is not considered in the dynamic model of the slave robot. It is noted that this error can be reduced by adopting gravity compensator integrated with the controller. Now, the cutting surgery is undertaken by adopting the tissue as a piece of pork in which the intact tissue is partially dyed in black color to mark the tumor part on it. Fig. 10 shows the photograph of the original and cut tissues achieved under two different conditions: without the haptic feedback and with the haptic feedback control action. It is noted the surgery without the haptic feedback is undertaken by directly observing the cutting tissue without any sensation information from the haptic master. It is clearly observed that more accurate cutting surgery is obtained with the haptic feedback control as expected. The preliminary surgical results presented in this work are quite self-explanatory justifying that the proposed method can possibly apply to real robot-assisted minimally invasive surgery (MIS) after augmenting more sophisticated control logics and teleoperated function. 6. Conclusions In this work, a new type of the haptic master which is applicable to the robot-assisted cutting surgery was devised using ER actuators and its effectiveness was validated by experimental tests. The proposed haptic master and slave robot were designed and manufactured using the dynamic models of ER spherical and linear actuators and kinematic model of the slave robot. Prior to performing the cutting surgery, the effectiveness of the proposed
Fig. 9. Comparison of desired force and actual force occurred from the cutting surgery.
Y.-H. Hwang et al. / Sensors and Actuators A 249 (2016) 163–171
Fig. 10. Tumor removal with and without feedback control.
haptic master was proved showing excellent agreement between the desired and actual forces in all manipulation directions. For example, it has been identified that the average force error does not exceed 0.038 N, and the standard deviation is 0.022 N. This accurate results provide the information how the operator precisely manipulates the slave robot based on the haptic sensation feedback signal (repulsive force). The proposed haptic master was connected to the slave robot in order to realize a simple surgery: cutting a pork. In order to observe the difference between with and without the haptic feedback signal, the part of the tissue (pork) was indexed by black ink. It has been demonstrated by experiment that more accurate cutting surgery can be achieved with the haptic feedback signal as expected. This indicates that the proposed haptic master featuring simple ER actuators can be one of potential candidates for the robot-assisted MIS in medical applications. It is noted here that the proposed ER haptic master doesn’t have any motion limitation. In addition, the durability of accuracy depends on the characteristics of ER fluid. It has been proved that there is no change of material characteristics of ER fluid under temperature of 100 ◦ C and the durability is guaranteed up to 106 cycles. However, a caution during the operation is required since a high voltage (about 2 kV) needs to be applied to get required force and torque during the surgery. Normally, the input current (about 2 mA) is very low, and hence the power to get the required force or torque is less than 5W for the surgical operation. It should be also remarked here that this work doesn’t contain how to control force feedback for the teleoperation which is crucial issue for final commercialization. The study on the distance-based teleoperation for the force feedback control of the proposed haptic master will be carried out in the near future as a second phase of this work. Acknowledgement This work was supported by National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (No. 2015R1A2A1A5054000). This financial support is gratefully acknowledged. References [1] Intuitive surgical, http://www.intuitivesurgical.com. [2] V. Hayward, O.R. Astley, M. Cruz-Hernandez, D. Grant, G. Robles-De-La-Torre, Haptic interfaces and devices, Sens. Rev. 24 (2004) 16–29. [3] J. Aoki, T. Murakami, A method of road condition estimation and feedback utilizing haptic pedal Advanced Motion Control, 2008. AMC’08. 10th IEEE International Workshop on, IEEE, 2008 pp. 777–782.
171
[4] A. Frisoli, E. Sotgiu, C. Avizzano, D. Checcacci, M. Bergamasco, Force-based impedance control of a haptic master system for teleoperation, Sens. Rev. 24 (2004) 42–50. [5] M. MacFarlane, J. Rosen, B. Hannaford, C. Pellegrini, M. Sinanan, Force-feedback grasper helps restore sense of touch in minimally invasive surgery, J. Gastrointest. Surg. 3 (1999) 278–285. [6] D. Ryu, C. Cho, M. Kim, J.-B. Song, Design of a 6 DOF haptic master for teleoperation of a mobile manipulator, Robotics and Automation, 2003. Proceedings. ICRA’03. IEEE International Conference on, IEEE, (2003) 3243–3248. [7] N. Diolaiti, C. Melchiorri, Teleoperation of a mobile robot through haptic feedback, Haptic Virtual Environments and Their Applications, IEEE International Workshop 2002 HAVE, IEEE (2002) 67–72. [8] F. Pierrot, E. Dombre, E. Dégoulange, L. Urbain, P. Caron, S. Boudet, J. Gariépy, J.-L. Mégnien, Hippocrate: a safe robot arm for medical applications with force feedback, Med. Image Anal. 3 (1999) 285–300. [9] S. Do Kweon, I.O. Park, Y.H. Son, J. Choi, H.Y. Oh, Linear vibration motor using resonance frequency, Google Patents, 2008. [10] A. Baz, K. Imam, J. McCoy, Active vibration control of flexible beams using shape memory actuators, J. Sound Vib. 140 (1990) 437–456. [11] S.-S. Han, S.-B. Choi, C.-C. Cheong, Position control of X–Y table mechanism using electro-rheological clutches, Mech. Mach. Theory 35 (2000) 1563–1577. [12] W. Li, B. Liu, P.B. Kosasih, X. Zhang, A 2-DOF MR actuator joystick for virtual reality applications, Sens. Actuators A: Phys. 137 (2007) 308–320. [13] D. Senkal, H. Gurocak, Spherical brake with MR fluid as multi degree of freedom actuator for haptics, J. Intell. Mater. Syst. Struct. 20 (2009) 2149–2160. [14] J. Blake, H.B. Gurocak, Haptic glove with MR brakes for virtual reality, Mechatron. IEEE/ASME Trans. 14 (2009) 606–615. [15] T. Kikuchi, K. Fukushima, J. Furusho, T. Ozawa, Development of Quasi-3DOF upper limb rehabilitation system using ER brake: PLEMO-P1, J. Phys.: Conf. Ser. (2009) 012015, IOP Publishing. [16] C. Mavroidis, C. Pfeiffer, J. Celestino, Y. Bar-Cohen, Design and modeling of an electro-rheological fluid based haptic interface, Proceedings of the 2000 ASME Mechanisms and Robotics Conference (2000) 10–13. [17] H. Böse, M. Baumann, G. Monkman, S. Egersdörfer, A. Tunayar, H. Freimuth, H. Ermert, W. Khaled, A new ER fluid based haptic actuator system for virtual reality, Int. J. Mod. Phys. B 19 (2005) 1628–1634. [18] F. Ahmadkhanlou, G. Washington, S.E. Bechtel, Modeling and control of single and two degree of freedom magnetorheological fluid-Based haptic systems for telerobotic surgery, J. Intell. Mater. Syst. Struct. (2009). [19] J. An, D.-S. Kwon, Modeling of a magnetorheological actuator including magnetic hysteresis, J. Intell. Mater. Syst. Struct. 14 (9) (2003) 541–550. [20] P. Bachman, A. Milecki, MR haptic joystick in control of virtual servo drive, J. Phys.: Conf. Ser. 149 (1) (2009) 012034. [21] J. Blake, H.B. Gurocak, Haptic glove with MR brakes for virtual reality, Mechatron. IEEE/ASME Trans. 14 (5) (2009) 606–615. [22] A.M. Okamura, Methods for haptic feedback in teleoperated robot-assisted surgery, Ind. Robot: Int. J. 31 (6) (2004) 499–508. [23] Reiley, Takintope Akinbiyi, Darius Burschka, David C. Chang, Allison M. Okamura, David D. Yuh, Effects of visual force feedback on robot-assisted surgical task performance, J. Thorac. Cardiovasc. Surg. 135 (2008) 196–202. [24] E.P. Scilingo, A. Bicchi, D. De-Rossi, A. Scotto, A magnetorheological fluid as a haptic display to replicate perceived biological tissues compliance, Microtechnologies in Medicine and Biology, 1st Annual International, Conference On 2000 (2000) 229–233. [25] P.H. Kim, S.M. Kim, Y.D. Park, S.B. Choi, Force modeling for incersions into various tissues with MRF haptic master, Smart Mater. Struct. 25 (3) (2016) 13, Article No. 035008. [26] S. Choi, D. Lee, Rotational motion control of a washing machine using electrorheological clutches and brakes, Proc. Inst. Mech. Eng. Part C: J. Mech. Eng. Sci. 219 (2005) 627–637. [27] S.G. Lim, J.S. Oh, P.B. Nguyen, S.B. Choi, A medical haptic master featuring bi-directional clutch and brake using electrorheological fluid, Adv. Sci. Lett. 13 (1) (2012) 322–326. [28] Q.-H. Nguyen, S.-B. Choi, A new approach for dynamic modeling of an electrorheological damper using a lumped parameter method, Smart Mater. Struct. 18 (2009) 115020.
Biography
Seung-Bok Choi is a Professor of Inha University and Director of Smart Structures and Systems Laboratory. He received his B.S. degree in Mechanical engineering from Inha University, Korea in 1979, and his M.S. and Ph.D. degrees in Mechanical engineering from Michigan State University, USA in 1986 and 1990 respectively. His research is focused on the applications of magnetorheolgocial, electrorheological fluids, piezoelectric materials and shape memory alloys.