Professional Documents
Culture Documents
0F
Authorized licensed use limited to: Sejong Univ. Downloaded on February 06,2023 at 12:00:10 UTC from IEEE Xplore. Restrictions apply.
cooperative platform could not offer motion scaling, which is
beneficial for fine manipulation in microsurgery [11].
As an alternative, a handheld type of miniaturized surgical
robots has been developed for retinal microsurgery and cell
manipulation [11]–[13]. It allows surgeons to directly
maneuver surgical tools, while selectively filtering out
erroneous motion such as hand tremor. Thus, surgeons can
attain the natural feel of manual operation and also direct
tactile feedback from the tool attached to the handheld
micromanipulator. The design of the handheld
micromanipulator in [12] has six degrees of freedom in
actuation and also offers an enlarged workspace in a level of
several millimeters at a tool tip. The use of six piezoelectric
linear motors allows the manipulator compact while providing
Figure 2. Kinematic configuration of a 6-PUS parallel mechanism with the
a cylindrical workspace with ⌀4 mm × 4 mm [12]. However, tool tip displaced by (-1.4, 1.2, 1.0), where the RCM is fixed.
the 6-DOF handheld micromanipulator still entails several
shortcomings in terms of insufficient force capability, inherent
are fixed at a base frame. It thus yields higher dynamic
fragility, and the limited dynamic performance. The tiny
performance, and collision between moving links is avoided
actuators used can withstand stall force only up to 0.5 N,
[14].
which yields a 0.25-N side load capability at the RCM. In
addition, the flexures made of polypropylene are prone to B. Inverse Kinematics
failure for undesired external force applied accidently. Inverse kinematics is adopted for the design and control of
Furthermore, the limited slew rate of the actuator and the 6-PUS parallel mechanism. The inverse kinematics
insufficient structural stiffness by use of the flexures determines the displacements of the prismatic joints for the
consequently degrade the overall dynamic performance of the target pose of an end-effector given an RCM. Such a pivoting
micromanipulator. Another handheld micromanipulator in motion is generally required in most of minimally invasive
[13] adopts a flexure based structure to reduce clearance surgery in order to prevent undesired transverse movement at
caused by conventional universal and spherical joints. the point of entry [15].
However, its workspace is limited to a hundred micrometer
due to the flexure joints. Given the geometric design parameters of the
micromanipulator, its nominal configuration is defined as in
In this paper, we propose a new fully handheld (1),
microsurgical robot that provides the higher force capability
and structural stiffness with improved dynamic performance Dt cos A Dt T
Authorized licensed use limited to: Sejong Univ. Downloaded on February 06,2023 at 12:00:10 UTC from IEEE Xplore. Restrictions apply.
Given the tool tip position, Te and the RCM, Tr , the generalized force acting on each actuation module, as
relative displacements from P0 are then described by the yielding the optimal geometric parameters, such as the
diameters of the base and moving platform and the length of
position vectors P0Te and P0Tr , respectively, as in (3).
the connecting link.
T
P0Te Tex Tey Tez lt The first step for the optimization is to calculate the force
(3) acting on the actuators given the specific amount and direction
T
P0Tr Trx Try lrcm , of the side load. We evaluate the generalized force on the
actuators by taking the Plücker coordinate representation. The
where lt and lrcm are the distance of the tool tip and the RCM unit screw ei is formulated by cascading two vectors, si and
from the center of the moving platform. Given a tool vector
si 0 , where si is the normalized vector of ith link and si 0 is
vtool , the center of the moving platform OP is defined as in
the cross product of PAi and the normalized vector of ith link
(4).
(8).
OP OP0 P0Te lt vtool , (4) T
T Tr M AT PA M A T
where vtool e . ei si si 0 i i i
61 i i
(8)
Te Tr M i Ai M i Ai
As a result, we describe the transformation of the moving
platform as the transformation of the tool vector from the From the unit screw ei , the Plücker coordinate, E , can be
nominal pose. The orientation of the transformation is constructed for the given pose of the tool, as in (9).
represented by the angle-axis rotation of the tool vector. The
axis of rotation vrot is the cross product of initial vector of the E 66 e1 e2 e3 e4 e5 e6 (9)
tool vinit and the tool vector vtool in (4). The rotation angle The generalized force on the linear actuators, f , is then
θrot is an angle between the two tool vectors, vinit and vtool calculated by equations from force and moment equilibrium
(5). (10).
f GE 1F ,
vinit vtool
vrot , θrot cos 1 (
T
vinit vtool (10)
) (5) where f 61 f1 f6
T
f2 f3 f4 f5
vinit vtool vinit vtool
T
Given the axis and the angle, the unit quaternion, q is defined F 61 F x Fy Fz Mx My M z
as in (6).
M i Aiz
G 66 diag ( ).
q cos rot sin rot vrot (6) M i Ai
2 2
F 61 is a force and moment vector generalized on the moving
Once the rotation matrix Rquat is formulated from the unit
platform, when the external force is applied to the tool. The
quaternion, the ith joint at the moving platform OAi is G 66 is a diagonal matrix, where each diagonal component is
determined by (7). the z-coordinate from the unit vector of Mi Ai .
OAi Rquat P0 Ai OP (7)
The generalized forces on the actuators are affected
z primarily by the geometric parameters, such as the diameters
Finally, we obtain the desired travel of the ith actuator, OM , i
of the base Db , and the moving platform Dt , and the length
by substituting (7) into (2).
of the links ls , since the unit screw ei in (8) is a function of
C. Geometric Parameter Optimization Db , Dt , and ls . While the previous study in [12] took a cost
In microsurgery, a small level of force is delivered to the function as the maximum force-variation among the
tool tip for delicate operation, while a relatively large force piezoelectric linear motors because of unavoidable preloading
may occur on the RCM. For example, the surgeons for usage of the motor, we define an objective function subject
deliberately move the eyeball while manipulating surgical to minimizing the maximum force distributed on the linear
tools in vitreoretinal surgery to expand the small field of view actuation modules, given the amount Fr and the direction r
obtained through an operating microscope with a high of the side load, respectively (11).
magnification. In this scenario, major external force would be
applied to the RCM, and is consequently distributed on each
link. Therefore, it is crucial to design a manipulator in which Db , Dt , ls
arg min f max Db , Dt , ls , Fr , r Fr 3.0 N , r 0, 2 (11)
the force distribution is well below the stall force of a linear
actuation module used for seamless operation. Our goal is to Furthermore, we also set a desired cylindrical workspace in
draw the optimal design parameters of the handheld terms of the diameter of workspace Dc and the height hc as a
micromanipulator, such that minimizes the force acting on the nonlinear constraint during optimization. Finally, the
actuator given side load applied to the RCM. optimization results in the diameters of the base and the
Since small actuators and mechanical components are moving platform and the link length, which can minimize the
susceptible to external force, this procedure is to minimize applied force on the linear actuators given the desired
workspace and the external force at the RCM.
12291
Authorized licensed use limited to: Sejong Univ. Downloaded on February 06,2023 at 12:00:10 UTC from IEEE Xplore. Restrictions apply.
TABLE I. OPTIMIZATION RESULTS
**
when a side load of 3.0 N is applied to the RCM.
12292
Authorized licensed use limited to: Sejong Univ. Downloaded on February 06,2023 at 12:00:10 UTC from IEEE Xplore. Restrictions apply.
A handpiece encloses the manipulator with PCBs (printed Host PC
Main
Controller RS-422
Handheld
Robot
circuit boards). The handpiece also includes a guide shaft with MFC Teensy 4.0 Viper EM Tracker
a 5-mm through hole to offer a path for insertion of surgical Data Acquisition
Controller
tools. Finally, an electromagnetic (EM) tracker sensor (Viper Plot
the handpiece. The EM tracker provides the 6-DOF pose of Command Coordinate Viper
Source
Transformation
the handpiece for controlling the robot in handheld operation. Filter Design
BLDC
Motors
The overall dimension of the handheld robot and manipulator Coordinate
Kinematics
coordinates and the ith (i = 1, 2, 3) fiducial point at the IV. EXPERIMENTS AND RESULTS
moving platform coordinates, respectively. Given the known First, we investigated the static accuracy of the EM tracker
position of the fiducial markers at the manipulator with respect to the distance between the EM tracker and the
coordinates, we run is a nonlinear optimization using the source. Because the pose of the tracker is used as the ground
S
‘fmincon’ function in Matlab™ to find H M which describe truth signal for all of handheld operation, it should precisely be
managed during microsurgical operation. As a result, the
a homogenous transformation from the manipulator, {M} to root-mean-square-error (RMSE) of the tracker was below 10
the EM tracker, {S} coordinates, where H S is defined by
W µm within in a range of 150 mm. The RMSE was still
maintained by a level below 20 µm up to a 250 mm-range.
the 6-DOF pose of the EM tracker. As a result, we obtain the
W W We evaluated the force capability of the handheld robot by
null position of the tool tip and RCM, X tip and X rcm , measuring the RMSE between the target and the resulting
described at the world coordinates using (14). individual motor position. Indeed, the RMSE of the tool tip
position should be measured regarding to external force
W
X tip H SW H MS X rcm
W
(14) applied, but the tool tip sensor is unavailable yet. Fig. 6 shows
the RMSE according to the side load for the range from 0 to
C. Control System 5.0 N on the RCM when a sinusoidal input of 1.0 mm was
A real-time controller was developed for the 6-DOF injected to the tip. As the side load increased, the RMSE
handheld micromanipulator using a 32-bit ARM Cortex-M7 gradually increased. The increment of RMSE for a 5.0 N side
processor with a 600 MHz clock (Teensy4.0, PJRC, USA). To load from zero-loading was only 2.2 µm, which corresponds
control the BLDC motors of the actuation modules, six slave to an error increment of 3.1 µm at the tip by simulation. As a
motor controllers (EPOS4, Maxon Motor Ag., Switzerland) result, the manipulator tolerates a side load up to 5.0 N without
communicate with a main controller via the CANopen significant control degradation, as expected from the
protocol. Given a specific pose of the end-effector, the target workspace analysis: up to 4.4 N at the nominal pose.
displacement of each actuation module is calculated via the
inverse kinematics and then an error between the target and Finally, we tested the performance of the active tremor
the null position is regulated by a feedback controller running cancellation as shown in Fig. 7 (a). The target task was to
at 500 Hz. The 6-DOF pose of the handpiece from the EM maintain contact force while holding the handheld robot over a
tracker is streamed to the main controller at 960 Hz via silicone membrane with a F/T sensor (ATI-Nano17, ATI
RS-422 communication as shown in Fig. 5. Industrial Automation, USA) on the top for 20 s under a
stereo-microscope (SZX7, Olympus, Japan). We computed
D. Active Tremor Cancellation the magnitude of peak forces for three-axes(xyz) from the F/T
Active tremor cancellation is to compensate for undesired sensor while executing the task for unaided (cancellation-off)
hand tremor while preserving voluntary motion. The and robot-aided (cancellation-on) operation. The task was
12293
Authorized licensed use limited to: Sejong Univ. Downloaded on February 06,2023 at 12:00:10 UTC from IEEE Xplore. Restrictions apply.
EM
Sensor
Silicone
EM
Source
Handhled
Force Robot
Sensor
(a)
Figure 6. RMSEs in actuator control for various sideloads applied to the
RCM.
V. DISCUSSION Figure 7. (a) Experimental setup for active tremor cancellation. (b)
Comparison of unided vs. robot-aided opeartion for task mainating contact
A new type of handheld robot was designed while taking force.
advantages of the 6-PUS parallel mechanism such as
inherently low inertia and the design flexibility. Compared to travel of the linear actuator and the reachable workspace.
the latest 6-DOF handheld micromanipulator [12], the new However, in the 6-PUS design, the travel of the actuator can be
design exhibits an order of magnitude higher force capability. easily customized by lengthening the lead screw regardless of
This is achieved primarily by the design flexibility, which the optimal geometric parameters. In addition, we also
allows to use larger and stronger actuators. In the previous revisited the optimization procedure of the handheld
design with the Stewart platform in [12], the use of larger micromanipulator introduced in [12] with an additional
actuators would fairly be limited because of the volumes of the nonlinear constraint to attain the desired workspace.
actuators used as extensible links between the moving and the
In addition, the proposed handheld robot that incorporates
base frames. On the other hand, in the 6-PUS platform, the
the highly accurate EM tracker, which is crucial for
larger actuator can be used because its body is located at the
identifying 6-DOF hand motion and for controlling tip motion.
base and stationary during manipulation. It thus offers a wide
A highly precise optical tracker has been introduced to address
range of actuator selection and more degrees-of-freedom in
issues raised by using the IMUs such as drift and low
design. Therefore, the proposed design can endure a larger
signal-to-noise-ratio (SNR) for relatively small and slow
amount of side load even with smaller base diameter, which is
signal. However, the optical tracker still suffers from a
a primary design parameter that determines the overall
line-of-sight problem, which hinders a large angular motion.
dimension of the handheld robot. As a result, the overall
The use of the EM tracker is free from such a sight-line issue
diameter of the base could be reduced from 20 mm to 16 mm,
and also not limited by the working distance from light
which is fairly significant reduction for the ergonomic grip of
detectors.
such handheld instruments. However, such a high force
capability would be achieved only if the mechanical joints can The performance of the handheld micromanipulator
endure the force delivered to the links. Although the flexure should more extensively be investigated in further study. The
design introduced in [12] less likely involves friction in current design of the handheld robot includes only one EM
movement and is easy to use, the flexure joint is prone to tracker to sense 6-DOF hand motion. By incorporating another
failure under high angular motion or high external load; if then, tracker that can identify the current pose of the end-effector tip,
the plastic deformation or irrecoverable damage would occur. we will be able to evaluate the overall performance of the
Hence, the new handheld robot adopting the custom-built manipulator in terms of positioning accuracy for various tasks
mechanical joints can endure a large amount of force without and control bandwidth. Besides, the use of the lowpass filter as
collapse of the manipulator. the active tremor cancellation inevitably imposes control
latency due to the inherent phase lag of the filter. It could also
The 6-PUS kinematic chain also offers another aspect of
be improved by applying feedforward control and
the design flexibility during the optimization procedure. In the
sophisticated filter algorithms to the handheld robot.
Stewart platform, the length of the extensible link is one of the
geometric parameters to be optimized for the minimal force
distribution, in which its specific length is coupled with the
12294
Authorized licensed use limited to: Sejong Univ. Downloaded on February 06,2023 at 12:00:10 UTC from IEEE Xplore. Restrictions apply.
REFERENCES
[1] H. F. Fine, W. Wei, R. Goldman, and N. Simaan, “Robot-assisted
ophthalmic surgery,” Canadian Journal of Ophthalmology, vol. 45, no.
6, pp. 581–584, 2010.
[2] S. M. Prasad, S. M. Prasad, H. S. Maniar, et al., “Surgical robotics:
Impact of motion scaling on task performance,” J. Am. Coll. Surg., vol.
199, no. 6, pp. 863–868, 2004.
[3] V. Swetha, E. Jeganathan, and S. Shah, “Robotic technology in
ophthalmic surgery,” Curr. Opin. Ophthalmol., vol. 21, no. 1, pp.
75–80, 2010.
[4] G. Ballantyne and F. Moll, “The da vinci telerobotic surgical system:
The virtual operative field and telepresence surgery,” Surg. Clin.
North Amer., vol. 83, no. 6, pp. 1293–1304, 2003.
[5] R. E. MacLaren, T. Edwards, K. Xue, et al., “Results from the first use
of a robot to operate inside the human eye,” Invest. Ophthalmol. Vis.
Sci., vol. 58, no. 8, p. 1185, Jun. 2017.
[6] M. Shoham, M. Burman, E. Zehavi, et al., “Bone-mounted miniature
robot for surgical procedures: concept and clinical applications,” IEEE
Trans. Robot. Automat., vol. 19, no. 5, pp. 893–901, 2003.
[7] M. L. Nordlund, D. M. V. Marques, F. F. Marques, et al., “Techniques
for managing common complications of cataract surgery,” Curr. Opin.
Ophthalmol., vol. 14, no. 1, pp. 7–19, 2003.
[8] R. H. Taylor, P. Jensen, L. Whitcomb, et al., “A steady-hand robotic
system for microsurgical augmentation,” Int. J. Robot. Res., vol. 18,
pp. 1201–1210, 1999.
[9] A. Uneri, M. A. Balicki, J. Handa, et al., “New steady-hand eye robot
with micro-force sensing for vitreoretinal surgery,” in Proc. IEEE RAS
EMBS Int. Conf. Biomed. Robot. Biomechatron., 2010, pp. 814–819
[10] A. Gijbels, J. Smits, L. Schoevaerdts, et al., “In-human robot-assisted
retinal vein cannulation, a world first,” Ann. Biomed. Eng., vol. 46, pp.
1676–1685, 2018.
[11] R. A. MacLachlan, B. C. Becker, J. C. Tabarés, et al., “Micron: An
actively stabilized handheld tool for microsurgery,” IEEE Trans.
Robot., vol. 28, no. 1, pp. 195–212, 2012.
[12] S. Yang, R. A. MacLachlan, and C. N. Riviere, “Manipulator design
and operation of a six-degree-of-freedom handheld tremor-canceling
microsurgical instrument,” IEEE/ASME Trans. Mechatronics, vol. 20,
no. 2, pp. 761–772, 2015.
[13] T. Zhang, L. Gong, S. Wang, et al., “Hand-held instrument with
integrated parallel mechanism for active tremor compensation during
microsurgery,” Ann. Biomed. Eng., vol. 48, no. 1, pp. 413–425, 2020.
[14] L. Zago and S. Droz, “A small parallel manipulator for the active
alignment and focusing of the secondary mirror of the VLTI ATS,” in
Proc. SPIE, 2000, vol. 4003, pp. 450–455.
[15] R. H. Taylor and D. Stoianovici, “Medical robotics in
computer-integrated surgery,” IEEE Trans. Robot. Autom., vol. 19, no.
5, pp. 765–781, 2003.
[16] C. Gosselin and J. Angeles, “Singularity Analysis of Closed-Loop
Kinematic Chains,” IEEE Trans. Robot. Automat., vol. 6, no. 3, pp.
281–290, 1990.
[17] C. N. Riviere and P. K. Khosla, “Characteristics of hand motion of eye
surgeons,” in Proc. IEEE 19th Annu. Int. Conf., vol. 4, Oct. 30–Nov. 2
1997, pp. 1690–1693. 199.
[18] S. Tatinati, K. C. Veluvolu, and W. T. Ang, “Multistep prediction of
physiological tremor based on machine learning for robotics assisted
microsurgery,” IEEE Trans. Cybern., vol. 45, no. 2, pp. 328–339, Feb.
2015.
12295
Authorized licensed use limited to: Sejong Univ. Downloaded on February 06,2023 at 12:00:10 UTC from IEEE Xplore. Restrictions apply.