`
`Mako Surgical Corp. Ex. 1013
`
`
`
`Iifllfi/Iijhiflflj/‘Jrfflflj
`
`“rrm‘u‘Jm/i‘m‘mJu ;
`
` \\
`
`I
`
`I
`
`012 2219
`
`_"
`
`Page 1
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`
`
`l
`
`The Use of Localizers, Robots and Synergistic Devices
`in CA8
`
`Jocelyne Troccaz * Michael Peshkin **
`
`Brian Davies W
`
`Ab stract
`
`There are many roles for electromechanical devices in image guided surgery.
`One is to help a surgeon accurately follow a preoperative plan. Devices for this
`purpose may be localizers, robots‘i, or recently, synergistic systems in which sur-
`geon and mechanism physically share control of the surgical tool. This paper
`discusses available technologies, and some emerging technologies, for guiding a
`surgical tool. Characteristics of each technology are discussed, and related to the
`needs which arise in surgical procedures. Three different approaches to syner—
`gistic systems, under study by the authors (PADyC, ACROBOT, and Cobots)=
`are highlighted.
`
`a"o-r.'.
`
`1
`
`Introduction
`
`An electromechanical device of some sort is needed in image—guided surgery,
`in order to connect the “information world” of images, plans, and computers,
`to the physical world of surgeons, patients, and tools. That is the situation in
`which a surgical plan has been created based on diagnostic images, and it is the
`job of the surgical system to guide the surgeon in the accurate execution of his
`own preoperative plan. The surgeon is again in direct contact with the surgical
`tool, but an interface device must also be connected to that tool, so that the
`computer may in some way provide guidance. Thought of as human interfaces,
`the perceptual quality of such a device is often the most prominent factor in
`the performance of surgical systems. We appreciate a quality that is sometimes
`called transparency — the quality of being perceptually absent. One purpose of
`this paper to describe the measures of interface device performance which deter—
`mine their suitability for use in various surgical situations. We give examples of
`surgical situations that particularly depend on one or another of these measures.
`Another purpose is to describe several classes of interface devices, with exam—
`ples. Previous descriptions of such devices relied on a decomposition in passive,
`active and semi—active systems [1] in which the degree of passivity was often
`associated with a type of technology. We prefer to define a new classification
`
`* TIMC/lMAG Laboratory, Faculté de Médecine de Grenoble, Domaine de la Merci,
`38706 La Tronche cedex - France
`** Mechanical Engineering Dept., Northwestern University, Evanston IL 60208 - USA
`*** Mechatronics in Medicine Lab., Imperial College of London, London SW7 2BX — UK
`4 By robot we mean a mechanism with some level of autonomy, programmability and
`adaptivity
`
`
`
`Mako Surgical Corp. Ex. 1013
`Page 2
`
`Page 2
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`728
`
`Moti
`
`joint
`so tl
`robo
`mecl
`into
`cont
`
`2-3
`
`
`
`based on function rather than mechanism including localizers, robots, and also
`a new class which we call synergistic devices. Synergistic devices are intended
`for direct physical guidance of a surgical tool which is also held and controlled
`directly by a surgeon. Each of the authors is pursuing a different approach to
`synergistic devices, and these approaches are outlined. The paper Concludes with
`a discussion of the applicability of the technologies to various surgical purposes.
`
`
`
`
`
`
`2 Classes of interface devices
`
`2.1 Localizers
`
`Localizers are devices that measure the coordinates of a tool or a pointer, but
`do not directly control that location. The location is controlled by the surgeon,
`by physically moving the tool or pointer, and is unconstrained by the localizer.
`Examples of localizers are passive arms with joint angle sensors, such as the Faro
`arm [2] Optical tracking systems perform a similar function, simply collecting .
`coordinates. Localizers have the advantage that achieving transparent behavior
`is easier than for devices with actuators. In other words they cooperate easily
`with a surgeon, interfering little with his intended motion. Lacking actuators,
`however, they cannot offer guidance to the surgeon by providing physical con-
`straint. Instead, the surgeon must explicitly observe and obey some other less
`immediate mode of guidance, usually a video display of some sort. An interesting
`variation is the addition of brakes to the joints of an otherwise unpowered local-
`'_
`izer arm. In this way, if a surgeon can be guided visually to position the arm in a.
`desired location according to a preoperative plan, the device can “lock” in that :
`position and can subsequently be used as a physical guide. However the intrinsic
`physical mode of the localizer is passive, allowing the surgeon full mobility.
`
`2.2 Robots
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Most robots are fully actuated, having a motor driving each joint. Thus the .__:'
`position of the robot’s end-effector is predominantly determined by how it runs
`its motors, and it intrinsically has little patience for physical “cooperation” wiLh
`a surgeon. For some applications no cooperation is required; the robot works i_
`autonomously. An example is the Adler/Latombe radiosurgery system, in which
`a heavy payload is moved about a large workspace, both of which exceed human _
`scale. No direct physical input from the surgeon is possible, or needed If see 54.2).
`'
`In some circumstances, the robot needs some help from the operator, for instance
`for registration. In this case the “cooperation” problem can be addressed by "
`adding a force sensor to the robot end—effector. The control computer is than
`aware of forces reflecting a surgeon’s intended motion. It may direct the robot
`motors to comply with that intent. In practice it has so far been very difficult to '
`acheive perceptually smooth cooperative motion in this way, but even primitive
`“force following” by the robot is useful (see §4.2}. Another approach, however,
`leaves some of the joints of the robot unactuated but still equipped with sensors.
`
`_
`
`J
`
`i'
`
`.
`
`-
`
`‘-..,,
`1L-on
`
`Make Surgical Corp. Ex. 10133
`Page
`
`Page 3
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`_
`i rI:
`
`729
`
`=
`
`Motion of these joints is naturally free and smooth. Since the decision to leave a
`joint unactuated is permanent, clever kinematic design of the robot is required
`._'."50 that the resulting “free” motions remain the appropriate ones even as the
`:- robot’s configuration changes. An example of these mixed acturated/unactuated
`' mechanisms is CMI’s Aesop, which holds a laparoscope inserted through a trocar
`* into the body (cf. §4.2). The intrinsic physical mode of actuated robots is active
`i' controlling position, and cooperating physically is not their natural mode.
`
`Part of our purpose in this paper is to introduce the notion of synergetic devices,
`in contrast to localizers and autonomous robots. Synergistic devices are intended
`for cooperative physical interaction with a surgeon. Both the surgeon and the
`synergistic device hold the tool, apply forces to it and to each other, and impart
`motions. Under computer control, the synergistic device may allow the surgeon
`to have control of motion within a particular plane, While the device dictates
`motion perpendicular to that plane, for instance. As an example, suppose the
`surgeon and the synergistic device cooperatively hold a bone saw. The surgeon
`may maneuver the saw at will within the defined plane, cutting at any desired
`speed from any angle of approach", and avoiding anatomic structures that must
`not be damaged. At the same time, the synergistic device confines the blade of
`the saw to a defined plane based on a preoperative plan, so that the eventual
`resected surface is flat and corresponds to the plan. Arbitrarily shaped surfaces,
`with greater or fewer than two dimensions, can be defined based on preoperative
`plans, and enforced by the synergistic device. The surgeon is free to control the
`remaining degrees of freedom. Synergistic cooperation has the benefit that the
`robot can provide accurate, precise geometric motions whilst the surgeon holding
`the tool can feel the forces applied and modify them appropriately. It also has
`the psychological benefit that the surgeon is in direct control of the procedure.
`Several of us have realized the value of a synergistic control of motion. Several
`distinct approaches to achieving that goal will be described in §4.3.
`
`erature (see [3] for instance).
`
`1' 2.3 Synergistic devices
`
`3 Technical needs
`
`In the following we will use the term mechanism to refer to any of these three
`types of systems. The technical specifications of robotic guiding systems answer
`a triple requirement: the ability to assist the execution of a given clinical task
`by providing accurate and repeatable precise geometric motions with intricate
`paths and repetetive motions tirelessly; ease of use for a clinical operator“, and
`with maximum safety for both the medical staff and the patient. In the follow—
`ing paragraph we will focus on user—oriented characteristics and safety—oriented
`characteristics only. Task—oriented characteristics are defined in the robotics lit-
`
`Mako Surgical Corp. Ex. 1013
`Page 4
`
`Page 4
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`
`
`730
`
`
` be carefully
`3.1 Operator-oriented characteristics
`mechanism
`The transparency of the system characterizes how user-friendly is the displace-
`ment of the mechanism. This qualitative factor depends mostly on the kinematic
`architecture of the mechanism which may promote some directions of motion
`e.g. wrist rotations are generally much easier to move than, say, a “reach" ex-
`tension. Quantitative dexteritymeasurements fmanipulability, isotropy, ...} have
`been introduced in the robotic domain [4] to quantify the kinematic extent to
`which a manipulator can attain all velocities. For instance, when a mechanism is
`isotropic no direction of displacement is favoured. Transparency also depends on
`the mechanical architecture of the mechanism and how light and well—balanced
`it appears to the operator. A compromise is often required between rigid struc-
`tures which can apply large forces over a wide region and the needs of sensitive=
`low inertia systems capable of high speed and accuracy.
`In the case of synergistic systems, the mechanism has to provide position
`and force information to make the operator feel the modelled data (anatomic
`obstacles for instance) and/or the task to be performed. Let us call their: con-
`straints. This results in several characteristics related to the type of constraints
`that can be provided by the mechanism. This is a design, control and modelling
`problem. As we will see further some systems give the feeling of rigid constraints
`whilst others feel deformable . In the former case, motions are completely forbid—
`den in certain directions. In the latter case, they are only resisted. Deformable
`constraints may be plastic or elastic. A plastic deformable constraint could be
`useful to make the operator feel obstacles from a certain distance. An elastic
`constraint would make the mechanism move on its own if the operator releases
`the end—effector. In other words the elastic would react whilst the plastic only
`resists the motions of the operator. The shape of available constraints is also of
`importance: it corresponds to the type of objects and tasks that can be felt. For
`instance, tasks such as linear drilling trajectories, planar cuts or 3D osteotomies
`for prosthesis implantation should be available to the orthopaedic operator. More
`complex constraints could combine position and orientation constraints. For in—
`stance, the tool could be constrained to follow accurately a. given trajectory with
`a given range of allowed orientations. Finally, one should know if constraints are
`programmable or not i.e. if they can be redefined for a new task or vary during
`the execution of one task. In addition to accuracy one should also characterize
`how smoothly the operator can move in free space or follow the border of the
`constraint. Frequency is an important factor linked to the interaction between
`the system and the operator. 200ms should be adequate to feel soft surfaces
`whilst Ims would be necessary to feel a hard surface.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`it is import
`conditions 1
`
`4 Tech]
`
`4.1 Loca
`
`Mechanical
`
`[5, 6] and f(
`anisms tha1
`
`position am
`is computec
`of the enco
`
`ranging fro
`forces appli
`compared ti
`being cumb
`track only 0
`bility of bei
`advantage i
`(however, i]
`when they i
`possible to
`when anthr‘
`
`the workspa
`have to be
`
`also a good
`operator ha
`as possible.
`intrinsically
`
`4.2 Robc
`
`ROBODO
`
`bones in hi]
`shape of the
`the implant
`term stabili
`
`
`
`
`3.2 Safety-oriented characteristics
`
`Reducing the maximum speed of the joints may increase safety. Adding redun-
`dancy to the mechanism (mechanics, control, sensors) may also improve safety.
`However this can also have the effect of allowing a large envelope, within which
`the position could be uncertain in the event of a failure. Failure modes have to
`
`workspace l'.
`bodoc syste
`and monito
`
`patient [11].
`guide the rc
`forms the SL
`
` Mako Surgical Corp. Ex. 1013
`
`Page 5
`
`Page 5
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`
`
`4 Technologies
`
`4.1 Localizers
`
`_
`
`'
`
`_
`
`'
`
`‘
`
`Is that can be felt. For
`.
`:nts or 3D osteotomies
`:paedic operator. More I” 3—'-_.
`on constraints. For in—“II *
`Lgiven trajectory with r {it
`:now if constraints are
`5'
`
`
`1"
`
`
`
`731
`
`
`be carefully studied for all types of mechanisms. For instance, the position of
`
`,1":
`.
`mechanism should keep stable when power 18 removed. For synergistic systems,
`
`it is important to know if the constraints can be violated and under which force
`355951135 is thedisp-lekie:
`
`
`gstlllygo'n the kinematic
`conditions they will.
`
`iefigrrectionsio _
`cit—En-
`
`53“
`.
`Etna.
`.“‘ -
`.-
`‘i-l‘.
`
`its 'i'sonspy'
`
`r‘__2I-lnn'ematic extflit: :3 .
`:-
`
`fil‘when a mechanism is'l
`I
`
`fitrency also depends on '
`
`'i!'g'ht and well—balanced-
`
`5d between rigid .struc
`
`._the needs of Sensitive
`
` as to provide ‘pOS'l 10‘
`
`sen-ed data (anatbin'i’éf-r,
`
`Iii-let us call them Iced:-
`the type of constraints-‘3'. —'.
`
`control and modelling; I
`
`ing of rigid constraints t
`are completely forbid—I
`3:: '
`
`
`
`y resisted. Deformable:
`' ..
`
`”ll ‘-:-"
`'
`
`ile constraint could beJ-l'
`.
`-
`.
`-;
`II
`'
`
`n dlstance. An elastic
`
`ft}
`"*1 -
`.1e operator releases, _
`whilst the plastic only'
`1'.
`0 constraints is also of 3'3-
`
`Mechanical localizers were introduced for endo—nasal surgery in the early eighties
`[-5. 6] and for neurosurgery W, 8]. They generally consist of man~powered mech~
`anisms that have several degrees of freedom and encoders on each joint. The
`position and orientation of a tool attached to the end—effector of the mechanism
`is computed in real— time from the geometric model and the instantaneous values
`of the encoders. In small workspaces that are typical in surgery, an accuracy
`ranging from 0.1mm to 1mm can be achieved. However, constraints or large
`Forces applied to the mechanism can deteriorate these values significantly. As
`compared to non mechanical navigators, these systems have the disadvantage of
`being cumbersome in the operative field. A major limitation is also that they can
`track only one object. How-ever, they always give information, without any possi—
`hility of being obstructed, as can occur with the non—contact localize-rs. Another
`advantage is that they can be fixed in a definite position to hold an instrument
`(however, in some systems, the application of brakes can cause a small motion
`when they are operated). Such a mechanism has to be as light and balanced as
`possible to limit the efforts to be produced by the human operator especially
`when anthropomorphic mechanical architectures (“arms”) are used. Therefore,
`the workspace and inertia have to be small and the “drag” on the various joints
`have to be similar. Transparency has to be as good as possible. This includes
`also a good visual interaction since all the topographic information given to the
`operator has to be rendered on displays. They must be as fast and ergonomic
`as possible. Because motions are generally man—powered, such mechanisms are
`intrinsically very safe.
`
`
`
`4.2 Robots
`
`
`
`
`
`ROB OD O C The Robodoc system has been developped for machining of femoral
`bones in hip surgery [9, 10}. Accurately machining the bone according to the
`shape of the prosthesis to be implanted allows perfect fit between the cavity and
`the implant and is intended to provide best biomechanical behaviour and long
`term stability of the implant. The robot is a SCARA based architecture which
`workspace has relatively limited interaction with the surgical field. In the Ho—
`bodoc system, the robot control subsystem performs an extensive safety check
`and monitors cutting force to ensure that unnecessary force cannot harm the
`patient [11]. The RoboDoc system uses force following to allow the surgeon to
`guide the robot into proximity of the surgical site, after which the robot per~
`forms the surgery autonomously.
`
`w task or vary during-s ',
`ould also characterize
`low the border of the’" i,
`0 interaction between
`3
`to feel soft surfaces
`
`.--
`5'..-
`
`','.'tJ-i"LOJJ'L
`
`
`
`
`
`
`
`
`safety. Adding redun-
`
`y also improve safety.
`
`
`ivolope, within which
`
`"hill/re modes have to
`”a
`
`
`
`
` Mako Surgical Corp. Ex. 1013
`
`Page 6
`
`Page 6
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`
`
`
`
`732
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`th
`
`Radiotherapy irradiation robot [12] developed a frameless system for neu—
`roradiosurgery based on the use of an industrial six axes robot which carries the
`radiation device. The robot is rather big because it has to position very accu-
`rately a heavy payload. It has parallel elements in its structure to increase its
`rigidity. Its very large workspace intersects the patient area. This system allows
`position tracking of the patient head during irradiation. Tracking is allowed for
`small motions only to avoid potential collision with the environment. A spherical
`architecture would have certainly been more adapted to this kind of application.
`A number of watch dogs are used at the control level to increase safety.
`AESOP The AESOP system from Computer Motion Inc. is dedicated to la-
`paroscopic procedures [13]. It is used to move the laparoscope and is controlled
`by foot pedals operated by the surgeon from video images. Mounted onto the
`surgical table, this SCARA-based architecture has a very limited workspace and
`a task—dedicated design. Indeed, it has 6 dofs, 4 actuated joints and 2 passive
`joints. The passive joints (no 4 and 5) are designed such that the laparoscope can
`rotate freely about the pivot point constraint imposed by the patient’s abominal
`wall therefore describing only conical motions. This provides the system with
`very interesting safety characteristics.
`
`4.3 Synergistic systems
`
`Measures of perceptual quality for synergistic devices might focus on the ques—
`tion of transparency as described in section 2. In particular we may ask how
`unobtrusive the device can be when it wishes to allow the surgeon full control
`over position. We will call this its transparency in “free mode”. Equally impor-
`tant is the smoothness with which the device can enforce a constraint surface.
`Optimally the surgeon would be able to use a softwaredefined constraint sur-
`face, as exhibited physically by a synergistic device, in much the same way that a
`surgeon normally uses a physical guide or jig. One wishes a guide to be smooth,
`preferably of low friction so that one may glide across it, and for it to be rigid
`and strong. We will refer to these characteristics as the smoothness of the device
`in “constraint mode”.
`Mechanical guide At the border of this classification, we can find systems
`such as [14] for which a six sures actuated mechanism autonomously positions a
`mechanical guide in stereotactic neurosurgery. This guide is used by the surgeon
`to guide a linear tool according a pre—planned trajectory. In this case, the con—
`straint is simple (a linear trajectory) and rigid.
`Moderated braking We mentioned above the possibility of superimposing
`brakes on the joints of a passive localizer arm. The surgeon thus has complete
`and free control of position, until a desired position is reached, at which point
`the localizer can be entirely ”frozen” and subsequently used as a rigid guide.
`An extension of this idea, which approaches the function of a synergistic device,
`is to use brakes which can be fractionally activated rather than turned entirely
`on or off. One would hope that, with appropriate control of the brakes, such a
`device could constrain the motion of a surgical tool grasped jointly by the robot
`and the surgeon, and could for instance confine the surgeon’s motion of the tool
`to with a region or on a plane. Such a device has in fact been explored in the
`CAS field by Taylor [15]. The transparency of such a device in its free mode can
`
`
`
`Mako Surgical Corp. Ex. 1013
`Page 7
`
`
`
`Page 7
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`733
`
`
`
`[gas-System for neu-
`"which carries the
`
`ition very accu-
`L
`ttfiire to increase its
`EThis system allows
`icki-ng is allowed for
`Briment. A spherical
`kind of application.
`fease safety.
`'is dedicated to la-
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`be excellent, since it reverts to being a passive Iocalizer when the brakes are off,
`and thus can be moved very easily. Unfortunately it turns out that brakes are
`
`extremely difficult to control smoothly. It is very difficult using brakes to exhibit
`
`a constraint surface at all, except in the fortuitous instances when one joint can
`
`be fully locked and another left fully free.
`
`Passive Arm with Dynamic Constraints (PADyC) Exhibiting a smooth
`
`constraint surface requires the establishment of allowed, non zero, velocities, such
`
`that the end—effector can move parallel to the constraint surface freely. Such a
`mechanism is the core of PADyC. In PADyC, each joint is velocity—limited by
`
`two reference Clutch plates, which thereby define an angular—velocity “window”.
`
`The joint may turn only at an angular velocity which falls between the limits
`established by the two reference clutch plates. The angular velocities of these two
`
`reference plates are controlled by a computer. When PADyC is in free mode, the
`
`angular velocity windows of all joints are set wide open, and the device allows
`unrestricted freedom of motion naturally. As a constraint surface is approached
`
`however, the angular velocity windows are made narrower in some directions,
`
`such that ultimately the only velocities available to each joint are the one which
`
`move the device away from or parallel to the constraint surface. Constraints
`include “free”, “position”, “trajectories” and “regions” modes [16]. Those are
`rigid and programmable constraints; nevertheless, some soft surface behaviours
`can be simulated by suitable velocity windowing. Because of its principle (no
`anticipation of next motion is made possible), PADyC natural modes are the
`free and region modes. The last one is particularly interesting for anatomic ob—
`stacle avoidance (neurosurgery or endonasal surgery for instance) and resection
`operations. The system is smooth and frictionless. A two link PADyC has been
`built (see figure 1a), and a three—link version is under construction.
`Cobots Armlike cobots with revolute joints are more difficult to describe
`than translational cobots, and interested readers are referred to [17, 18]. Suffice
`it to say that the principle of operation is the same. Here in the interest of space
`we will describe the simplest possible translational cobot. As presented below
`it is a two degree of freedom device. Several higher degree of freedom cobots
`are under development, as well as an armlike cobot. The two degree of freedom
`translational cobot consists of a rolling wheel, free to roll on a flat working sur—
`face. A computer controlled motor steers the wheel, and a handle is attached to
`it as shown in figure 1.b. The user and the cobot interact through this handle,
`and the workspace of the cobot is the horizontal plane in which the user can
`move the handle. Note that the motor cannot make the handle move; only the
`user can do that. The motor only steers. It can however enforce a constraint
`of superimposing
`surface, which in this example should be a understood as a constraint curve in
`thus has complete
`the planar workspace. It can enforce this constraint simply by steering the Wheel
`3d, at which point
`parallel to it. Because the rolling wheel can only be moved in the direction it is
`I as a rigid guide.
`
`aimed at each moment, the user perceives an impenatrable boundary at the con-
`synergistic device,
`
`straint surface. In practice this illusion is convincing. Since the constraint arises
`an turned entirely
`
`the brakes, such a
`mechanically, it is smooth and frictionless. The instrinsic modes of PADyC are
`
`)intly by the robot
`its free and region modes. In contrast, cobots have an instrinsic mode which is
`
`motion of the tool
`
`the trajectory mode. Free and region modes must be acheived through computer
`
`in explored in the
`control. To allow the user full freedom of motion, the control computer uses a
`its free mode can
`
`force sensor to detect which direction the user wishes to move the handle. It
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`9e and is controlled
`FMounted onto the
`i-i'ted workspace and
`oints and 2 passive
`the laparoscope can
`patient’s abominal
`es the system with
`
`focus on the ques-
`r we may ask how
`rurgeon full control
`'6”. Equally impor—
`constraint surface.
`ied constraint sur-
`
`he same way that a
`uide to be smooth,
`d for it to be rigid
`hness of the device
`
`: can find systems
`mously positions a
`sed by the surgeon
`this case, the con—
`
`
`
`
`
`
`
`
`Mako Surgical Corp. Ex. 1013
`Page 8
`
`Page 8
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`my:
`
`.03)
`
`
`Fig. 1. (a) PADyC: A two degrees of freedom laboratory prototype. The operator
`
`moves the PADyC in the plane and the system constrains the motion according to
`
`pre-planned strategy. (Courtesy of Dr J ocelyne Troccaz, TlMC Laboratory) (b) Cabot:
`
`A two degrees of freedom prototype. The operator moves the cobot in the plane using
`
`the handle and the system automatically rotates the wheel in order to describe a given
`
`trajectory. (Courtesy of Dr Michael Peshkin, North Western University) (c) Close—up of
`
`ACE 0B OT mechanism and end—effector showing controlled degrees of freedom (Cour—
`
`tesy of Dr Brian Davies, Imperial College of London)
`
`
`
`
`then steers the rolling wheel to coincide with the desired direction, much the
`
`way that a caster wheel under the leg of a piece of furniture aligns itself with
`
`the desired direction. The constraints are rigid and programmable.
`
`ACROBOT As mentioned in section 4.3, a conventional robot may attempt
`
`to cooperate with a human user by measuring the the user’s applied force and
`
`driving joint motors to comply with these forces. However the transparency is
`
`usually poor. ACROBOT (or Active Constraint ROBOT) is a robot specially de—
`
`signed for transparent cooperation with a human user, while nevertheless using
`
`actuated joints. ACROBOT uses backdrivable motors and transmissions, Where
`
`conventional robots are usually made strong and stiff at the expense of backdriv—
`
`ability. Mechanically, ACROBOT places the human user and the robot’s motor
`
`on a more nearly equal basis for controlling position. A conventional robot gives a
`
`great advantage to the robot’s motors, as it is intended to be insensitive to exter—
`
`nally applied forces. In ”region” mode ACROBOT’s motors are actively driven
`
`to comply with the user’s force. Good tranparency can result due to mechanical
`
`sharing of forces between motor and human, made possible by backdrivability.
`
`As the user approaches and then contacts a constraint surface defined in the
`
`preoperative plan, the motors are actuated to gradually increases its resistance
`
`until, at the edge of the permitted region, it prevents further motion by the op-
`
`erator. Constraints may be deformable (elastic or plastic) and rigid. Following
`
`preliminary trials of a prototype a new 4 axis robot, mounted on passive struc-
`
`ture, has been constructed and evaluated (see figure 1.c) Whilst the ACROBOT
`
`is currently being used for knee surgery, the system is also suited to a range of
`
`orthopaedic and soft tissue procedures.
`
`
`
`Mako Surgical Corp. Ex. 1013
`Page 9
`
`5 DiS(
`
`Synergisti
`must offer
`
`(based on
`that a ph)
`physical g
`the numb
`
`plan speci
`video guic
`ACROBO
`
`properties
`proaches 1
`Both of ti
`are the “rt
`
`tory” moc
`future of t
`Is the com
`and the su
`
`If so, local
`the surgeo
`surgical p1
`important
`put from t
`which the .
`
`surgeon a1
`tive and sc
`some aspe
`for others.
`
`Referen
`
`1. P. Cinc
`passive
`special
`. ISG Te
`
`[Q
`
`nologiei
`3. Craig J
`1986.
`'
`4. C. Kleii
`matical
`1987.
`v
`
`Page 9
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`...hJ-Jire-A'.—.
`
`
`“fa........_..._....._..-.......
`
`735
`
`5 Discussion
`
`Synergistic devices are intended to cooperate physically with a surgeon. They
`must offer good transparency, but also be able to produce forces. These forces
`(based on a preoperative plan) can guide the surgeon physically, much in the way
`that a physical jig or guide offers guidance. In contrast to video guidance, direct
`physical guidance promises to be more efficient and accurate. Especially when
`the number of degrees of freedom (angles and positions) that the preoperative
`plan specifies is large, it becomes difficult and frustrating for a surgeon to follow
`video guidance. Three approaches to synergistic devices were described. In one,
`ACROBOT, a special purpose robot is used. It may bring speed and tracking
`properties that PADyC and Cobots cannot. We also described two other ap-
`proaches to synergistic devices, both of which rely on novel joint mechanisms.
`Both of them are intrinsically safer than ACROBOT. PADyC’s intrinsic modes
`are the “region” and “free’7 modes, whilst Cobots’ intrinsic mode is the “trajec—
`tory” mode. ACROBOT has no preferred mode. It is debatable what will be the
`future of those CAS systems that perform surgery based on a preoperative plan.
`Is the computer’s primary role to be to present the plan Visually to the surgeon,
`and the surgeon works essentially freehand but with visual reference to the plan?
`If so, localizers offer the needed functionality with a minimum of interference to
`the surgeon’s delicate work. Or perhaps CAS systems will increasingly execute
`surgical plans themselves, with the surgeon’s direct touch becoming less and less
`important. If so, semi—autonomous robots with little opportunity for physical in-
`put from the surgeon may hold the future. Yet a third possibility, and the one to
`which the synergistic devices being developed by the authors is addressed, is that
`surgeon and computer will need to interact physically in a direct and coopera—
`tive and sensitive way. This is required if the surgeon is to remain responsible for
`some aspects of tool motion, while simultaneously the computer is responsible
`for others. Synergistic devices are designed for this intimate cooperation.
`
`-J
`
`References
`
`1. P. Cinquin and al. Computer Assisted Medical Interventions at TIMC Laboratory:
`passive and semi—active aids. IEEE Engineering in Medicine and Biology magazine,
`special issue Robots in Surgery, 14(3):254—263, 1995.
`ISG Technologies. Viewing wand operator’s guide. Technical report, ISG Tech—
`nologies lnc., 1993.
`
`.
`
`[O
`
`3. Craig John J.
`1986.
`
`Introduction to Robotics Mechanics and Control. Addison Wesley,
`
`4. C. Klein and B. Blaho. Dexterity Measures for the Design and Control of Kine—
`matically Redundant Manipulators.
`Int. J. Robotics Rea, 6(2):72—83, Summer
`1987.
`
`prototype. The operator
`the motion according to
`C Laboratory) (b) Cabot:
`cobot in the plane using
`order to describe a given
`'niversity) (c) Close—up of
`.egrees of freedom (Cour—
`
`ad direction, much the
`:iture aligns itself with
`rammable.
`
`LELI robot may attempt
`ser’s applied force and
`’er the transparency is
`is a robot specially de—
`hile nevertheless using
`:1 transmissions, where
`1e expense of backdriv—
`and the robot’s motor
`
`ventional robot gives a
`be insensitive to exter—
`
`ors are actively driven
`sult due to mechanical
`
`ble by backdrivability.
`surface defined in the
`ncreases its resistance
`
`her motion by the op—
`) and rigid. Following
`nted on passive struc—
`Vhilst the ACROBOT
`
`:0 suited to a range of
`
`
`
`
`
`Mako Surgical Corp. Ex. 1013
`Page 10
`
`Page 10
`
`Mako Surgical Corp. Ex. 1013
`
`
`
`
`
`
`
`736
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`gical navigation
`on Biomedical Engineering,
`
`In R.