Beyond the Manual Touch: Situational-aware Force Control for Increased Safety in Robot-assisted Skullbase Surgery

Bey ond the Man ual T ouc h: Situational-a w are F orce Con trol for Increased Safet y in Rob ot-assisted Skullbase Surgery Hisashi Ishida 1* , Deepa Galaiy a 1,2 , Nimesh Nagururu 2 , F rancis Creigh ton 1,2 , P eter Kazanzides 1 , Russell T a ylor 1,2 , Manish Sah u 1* 1 LCSR, Johns Hopkins Univ ersity , USA. 2 Departmen t of Otolaryngology , Johns Hopkins Universit y , USA. *Corresp onding author(s). E-mail(s): manish.sah u@jh u.edu ; Abstract Purp ose : Skullbase surgery demands exceptional precision when remo ving b one in the lateral skull base. Rob otic assistance can alleviate the effect of human sensory-motor limitations. How ever, the stiffness and inertia of the rob ot can significan tly impact the surgeon’s perception and con trol of the to ol-to-tissue in teraction forces. Metho ds : W e present a situational-aw are, force control technique aimed at reg- ulating interaction forces during rob ot-assisted skullbase drilling. The contextual in teraction information derived from the digital twin en vironment is used to enhance sensory p erception and suppress undesired high forces. Results : T o v alidate our approach, w e conducted initial feasibility exp eriments in volving a medical and tw o engineering students. The exp erimen t fo cused on further drilling around critical structures follo wing cortical mastoidectomy . The exp erimen t results demonstrate that robotic assistance coupled with our proposed con trol scheme effectively limited undesired interaction forces when compared to rob otic assistance without the prop osed force con trol. Conclusions : The prop osed force control techniques show promise in signif- ican tly reducing undesired interaction forces during robot-assisted skullbase surgery . These findings contribute to the ongoing efforts to enhance surgical precision and safet y in complex procedures inv olving the lateral skull base. Keyw ords: T ool-tissue interaction, Coop erative control, Adaptive force control, F orce feedback, Skullbase surgery , Surgical rob otics, Digital twin 1 1 In tro duction Skull base surgery is tec hnically c hallenging and requires precision in the execution of in tricate tas ks within the constrained and delicate lateral skull base environmen t. T o access this region, surgeons must skillfully drill through v arying densities of b one to rev eal critical anatomical structures, often hidden b y op erable tissue at sub-millimeter distances [ 1 , 2 ]. Microsurgery is constrained by the limits of h uman visual, tactile, and motor control thresholds [ 3 ], making the tissue manipulation a challenging endea vor. Consequen tly , surgeons undergo a rigorous minimum of 7 years [ 4 ] of p ostgraduate training to master the complexities of this anatomical region. One significant c hallenge lies in the subtle nature of to ol-to-tissue interaction forces, whic h can fall at or b elow human tactile p erception levels[ 5 ]. The introduction of rob otic assistance holds promise in enhancing surgical performance b y enhancing to ol tip precision and reducing the impact of hand tremors during delicate surgical tasks. Ho wev er, while robots can provide steady-hand manipulation, the sensory feedback traditionally derived from manual to ol in teractions is altered, p oten tially leading to unin tended con tact b etw een the instrument and tissue, with the p oten tial for serious injury . The accurate sensing and intelligen t control of tool-tissue interaction forces are piv otal in enhancing surgical safety and reducing the incidence of complications [ 6 , 7 ]. The success of rob otic assistance in skull base pro cedures hinges on the precision of the sensing and con trol mec hanisms that adapt to the surgical con text and ensure that to ol-tissue in teractions remain within safe thresholds. In this work, w e address these c hallenges through the dev elopment of a situation- a ware, adaptiv e con trol metho d tailored for controlled tissue ablation in skull base pro cedures. Our con trol method leverages contextual information from the dynami- cally c hanging surgical environmen t to regulate interaction forces concerning desired forces related to distinct anatomical structures. Giv en the tendency of medical trainees to apply more force than attending surgeons, con tributing to technical errors [ 8 ], we h yp othesize that our prop osed situational-aw are adaptive force control strategy can reduce undesired con tact forces during collaborative drilling in skull base pro cedures. T o v alidate our hypothesis, we conducted exp erimen ts with tw o attending sur- geons, performing an initial mastoidectomy around v arious anatomical structures. The resulting drilling v alues w ere statistically analyzed to determine exp ected forces during these pro cedures. Subsequen tly , safety threshold v alues for the force control scheme w ere established based on these statistical insights. T o demonstrate the effectiveness of our prop osed con trol scheme, w e conducted a cadav eric temp oral b one exp eriment in volving three inexp erienced users. The exp erimen t included drilling around critical structures follo wing cortical mastoidectomy . A comparative analysis of our prop osed metho d against co op erativ e robotic assistance without force control emplo yed a com- prehensiv e set of force metrics [ 8 ], including av erage forces, maximum forces, and the time sp en t ov er threshold force. The exp erimen tal results consistently demonstrate that the active robot control sc hemes maintain applied forces b elo w the desired safe threshold during rob ot-assisted skull base surgery . The con tributions of our work include: 1) the extension of the Digital Twin (DT) en vironment to incorp orate contact interactions through force measuremen ts, 2) the developmen t of an approach for accurate estimation of the currently op erated 2 anatomical structure based on contextual information in DT en vironment, and 3) the dev elopment of an adaptive force control technique for con trolled tissue ablation in skull base pro cedures. 2 Related W ork T raditional metho ds in rob ot-en vironment interaction are commonly classified in to t wo main categories: indirect force control (without explicit force feedback) and direct force control [ 9 , 10 ]. Indirect force control in volv es p erforming manipulation tasks by lev eraging the in terplay b et ween motion and in teraction forces. How ever, a p oten tial dra wback is the risk of losing contact with the environmen t due to the absence of explicit force feedback [ 11 ]. In contrast, direct force control facilitates manipulation b y explicitly tracking a desired force, making it particularly effective for surgical tasks where specific force profiles are essen tial. Ho w ever, the efficacy of direct force control is con tingent up on prior kno wledge of the surgical task and a situational understanding of the dynamic, time-v arying surgical environmen t. In the pursuit of enabling rob otic assistance in surgery , prior w ork [ 5 ] in the lit- erature has predominantly fo cused on adv ancing hardw are and sensor tec hnologies to facilitate steady-hand manipulation and reduce the impact of hand tremors. How ever, comparativ ely less attention has b een directed tow ard the dev elopment of control and sensing algorithms.[ 8 ] In the realm of orthop edic surgery , studies prop ose force con- trol approac hes to enhance p ositional accuracy of a hand-held rob ot [ 12 ]. Also, force sensing metho ds hav e b een developed for tasks such as screw-path-drilling [ 13 ] and co operatively assisted surgical needle insertion[ 14 ]. Additionally , a force-control-based approac h is dev elop ed for regulating penetration force during cell injection in zebrafish em bryos [ 15 ]. Within the context of ENT surgery , Rothbaum et al. [ 16 , 17 ] ev aluated the p oten tial of rob otic assistance in stap edotomy and show ed that it can significantly limit the maxim um force applied to the stap es footplate. Ebrahimi et al.[ 18 ] prop osed con trol metho ds for eye surgery to limit sclera forces applied to the to ol s haft at the trocar, further show casing the adaptabilit y of robotic systems in surgical con texts. Sang et. al. [ 19 ] integrated a force sensing drill to the end effector of a da Vinci rob ot and show ed its utility for rob otically assisted otologic surgery . Additionally , systems lik e MMS [ 20 ], utilizing a control console with t wo joystic ks, and RobOtol [ 21 ], whic h emplo ys a Phantom Omni jo ystick for force feedback, hav e b een explored. Ho wev er, these approac hes largely follow the teleop eration paradigm, introducing c hallenges suc h as removing surgeons from direct patient access and necessitating substantial alterations to conv en tional surgical procedures. In con trast, co operative systems ha ve emerged as a cost-effectiv e and minimally disruptiv e alternativ e. These systems offer in tuitive con trol ov er teleop eration, requiring less training for surgeons to b ecome profi- cien t. F urthermore, they enable surgeons to remain at the patien t’s bedside, enhancing the ov erall surgical exp erience. In our prior w ork [ 22 ], w e ha ve developed a force-sensing surgical drill for a hand-o ver-hand co op erativ ely controlled surgical robot that can measure to ol-to-tissue forces with millinewton accuracy in real-time. This paper extends our previous efforts, 3 fo cusing on the developmen t of adaptive force control schemes to activ ely constrain in teraction forces within the surgical environmen t. This inv olves seamless in tegration of patient anatom y , the co operative robot, and the interaction forces into a real-time DT environmen t as well as utilization of the contextual information in DT to dynami- cally adjust the control parameters. In a recen t study [ 23 ], we in tro duced safet y-driven virtual fixtures in tended to guide drill motion aw a y from critical structures. How- ev er, these fixtures did not account for the impact of to ol-tissue interaction forces. In contrast, this pap er places a distinct fo cus on the real-time in teraction forces, as w ell as the developmen t of situation-a ware active force control aimed at constraining in teractions with the surgical environmen t. 3 Metho ds In the follo wing section, w e describ e the prop osed situational-a ware force control approac h. The main motiv ation for this metho d is to balance high transparency dur- ing non-con tact p erio ds and ensure safer to ol-tissue interaction during contact. The prop osed metho d com bines the real-time en vironment information (to ol-tissue interac- tion force) and surgical context (op erating anatomical structure and safety margin for the structure) from the virtual simulator to enhance surgical safety . First, the impor- tance of com bining the environmen t information with surgical con text is describ ed in Sec. 3.1 . The co op erativ e rob otic system is desc ribed in Sec. 3.2 , and the main control sc heme is described in Sec. 3.3 . Fig. 1 System o verview sho wing force sensing drill, surgical en vironment, and digital twin. S denotes the op erating anatomical structure. x T is the drill tip p ose and ˙ q is a joint v elo cit y . F T , F W , F D denote to ol-tissue interaction force at the drill tip, and forces measured from the wrist force sensor and the drill sensor, resp ectiv ely . 4 3.1 Digital Twin with T o ol-tissue interactions The efficacy of any shared con trol functionalit y within the surgical con text necessitates a comprehensiv e understanding of the en vironment and the ongoing task. Lev eraging a DT approach holds the promise of deriving semantic knowledge ab out the surgical en vironment and the system’s state. Consequently , we in tegrated a DT environmen t tailored for skull base pro cedures [ 24 ]. The developmen t of the DT framework man- dates precise mo deling of individual system comp onen ts to closely em ulate a real surgical en vironmen t. Intraoperatively , these DT models undergo registration with the actual patient, enabling the tracking of surgical instrument p ositions relative to the patien t. How ever, due to the inherent inaccuracies of current tracking systems, rely- ing solely on spatial information prov es inadequate for precisely iden tifying contacts with the anatomy . In this work, we enhance the DT system by integrating force sens- ing measurements [ 22 ], crucial for establishing a real-time and precise understanding of to ol-tissue in teractions. This extension ensures a more nuanced and accurate rep- resen tation of the surgical scenario, ackno wledging the dynamic interpla y of spatial information and interaction forces. 3.2 Co op erativ e Rob ot The Rob otic ENT (Ear, Nose, and Throat) Microsurgery System (REMS) is a co operatively-con trolled rob ot sp ecialized for use within otolaryngology–head and nec k surgery [ 25 , 26 ]. In this study , we use a pre-clinical version developed by Galen Rob otics (Baltimore, MD). One of the primary adv antages of REMS is that it offers a significant b enefit in instrumen t stability in head and neck surgery . T he co operative mo de of the rob ot uses the admittance control [ 27 ] la w: arg min ∆ q ( | GF H − J ∆ q | ) (1) G ∈ R 6 × 6 denotes a diagonal matrix that represents the admittance gains, J ∈ R 6 × m is a Jacobian and ∆ q ∈ R m is a joint velocity vector. The incremental rob ot motion, ∆ x ∈ R 6 , is expressed as (∆ x = J ∆ q ). T o deliver a sensation of touc h and a void damaging the anatomy during contact without disturbing the surgeons’ op eration, we in tro duce a gain adjustment term, σ ∈ R . Consecutiv ely , the rob ot is con trolled b y the follo wing admittance la w. G ′ = σ I G is an adjusted admittance gain, where I ∈ R 6 × 6 is an identit y matrix. arg min ∆ q ( | G ′ F H − J ∆ q | ) (2) W e increase σ = σ hig h ( σ hig h > 1 . 0) so that the rob ot will resp ond more promptly to the h uman input when there is no con tact with the anatomy ( | F contact | < C ), where C is a contact threshold. Meanwhile, during the con tact, σ is set as σ contact ( σ contact < 1 . 0) to ensure precise and delicate motion. 3.3 Con trol sc hemes This section describ es the adaptiv e gain control scheme prop osed in this pap er. The bac k end mo dule inv olves the situation-aw are Digital Twin en vironment to identify 5 the op erating structure. The front end consists of adaptive force control to regulate the interaction b etw een the surgical and the patient anatom y . 3.3.1 Op erating structure estimation Giv en the delicate nature of each anatomical structure, precise regulation of interaction forces tailored to the specific anatomy is imp erativ e. The real-time computation of the minim um distance b et ween the drill tip and its nearest anatomical structure can b e acquired using signed distance fields [ 28 ]. Giv en n ∈ Z anatomical structures of the patien t mo del, one can determine the distance d n ∈ R b etw een the drill tip and the n th anatomical structure. The op erating structure, denoted as S , is identified based on the proximit y to the drill tip and real-time in teraction force measurements, which are contin uously track ed b y the digital twin paradigm describ ed in Section 3.1 . If the closest distance to an anatomical structure, d min ≤ ∀ d n , is less than a predefined threshold, γ n , and the force v alue surpass contact threshold, we recognize the closest structure as the op erating structure. Fig. 2 The c losest distances b et ween the drill tip and the anatomical structures, d n , are calculated in real-time. ( d 1 : F acial Nerve, d 2 : T egmen, d 3 : Sigmoid, d 4 : Cortical b one, d 5 : T rab ecular b one). F or eac h operating structure, a safety threshold denoted as λ n ∈ R is established as an acceptable interaction force. This allo ws the prop osed force con trol to incorp orate b oth real-time in teraction forces with the tissue and surgical context information. 3.3.2 Situational-aw are force con trol T o minimize the undesired force applied to the anatomy , we prop ose a situational- a ware force control method that adaptively adjusts the gain using b oth the to ol-tissue in teraction force and the operating structure. The con trol gains remain consistently high ( σ hig h ) to increase the transparency of the rob ot and adjusted when making con tact with the anatomy to enhance the sensation of touch. The gain adjustments are introduced during anatomical ablation. The adaptive gains remain unchanged ( σ contact ) until force v alues exceed safe force ranges ( C < | F contact | < U ). Once the force v alues surpass undesired ranges, the control metho d adjusts gains gradually , making the rob ot increasingly stiff (Eq. 3 ). This adjustment is prop ortional to the 6 o verall time sp en t b ey ond the desired force v alues. This gain adjustmen t term, σ ( t ) can b e expressed as σ ( t ) =      σ (0) exp ( − η ∆ F ( t − t 0 )) + σ low if | F contact |≥ U σ contact if U > | F contact |≥ C σ hig h if | F contact | < C (3) where U is an anatomy-specific threshold for undesired force deriv ed in Sec. 3.3.1 ( U = λ n ), and C is the contact threshold. σ (0) = σ contact − σ low and ∆ F = F contact − U . t 0 is the time when contact force, F contact , surpasses, U . η ∈ R represen ts a deca ying constan t. σ low ∈ R ( σ low < σ contact ) is a low er bound gain adjustmen t term that mitigates further anatomical con tact. This gain adjustment term is used in admittance con trol explained in Algorithm 1 , whic h hinders further damage to the tissue while main taining the con trol to follow the user input. arg min ∆ q ( | σ ( t ) I GF H − J ∆ q | ) (4) Algorithm 1 Situational-aw are force con trol Input: State of the environmen t: User input F H , T o ol-tissue interaction force F T , drill tip p ose w.r.t anatomy  x T Output: Rob ot motion ∆ q if | F T | > C then # Once there is a con tact S ← getOp eratingStructure(  x T ) # Get op erating structure U ← getInfo( S ) # Get undesired force threshold σ = σ contact while | F T | > U do # Once undesired force applied Calculate σ ( t ) based on equation 3 end while else # During freehand motion σ = σ hig h end if Calculate ∆ q with prop osed admittance control (equation 4 ) return ∆ q The o verall algorithm that summarizes the proposed metho d can b e found in Algo- rithm 1 . Since the op erating structure and its acceptable interaction force, λ n , are estimated from Section 3.3.1 , gain adjustmen t term, σ ( t ), changes its v alue in real time incorp orating the spatial information and the surgical context. Consequen tly , realiz- ing a situational-a ware adaptive force control strategy which can reduce undesired in teraction force during tissue ablation. 7 Fig. 3 Exp erimen tal setup. Medical student uses the surgical drill attached to the coop erativ e rob ot under microscopic view, while an adjacent optical track er monitors both the drill and the anatomy . 4 Exp erimen ts W e conducted a feasibility study to demonstrate our method’s application and to gather initial feedback for a larger user study in the future. 4.1 Exp erimen t setup In preparation for our exp erimen t, we attac hed registration pins to the cada veric temp oral b one. The b one was then scanned using high-resolution CBCT (Brainlab Lo opX, 0 . 26 mm 3 v oxel size); scans were then taken for each temp oral b one. Using 3D Slicer, w e annotated critical structures and pin lo cations. These annotations were instrumen tal in crafting the patient’s anatomical model for our virtual sim ulator. Prior to the exp erimen t, the temp oral b one was securely mounted in a designated holder, whic h was then anchored to a surgical table to eliminate uninten tional mov emen t. T o track mov ements during the exp erimen t, mark ers were placed on the surgical drill and the b one holder. F ollowing this, we used a 2 mm drill tip for pivot calibration, ac hieving an RMSE v alue of 0.1 mm . Finally , a p oin t-set registration metho d was p erformed to register the temp oral b one to the virtual simulation. These p oin ts were carefully sampled using the surgical drill, leading to an RMSE of less than 0.5 mm . During the exp erimen t, drilling pro cedures was conducted using a surgical microscop e (Haag-Streit, K¨ oniz, Switzerland) equipp ed with stereo vision (Fig. 3 ). F orce sensors are sampled in 200 H z , collaborative rob ot and virtual sim ulation are running in 500 H z and 1000 H z resp ectiv ely . 4.2 Exp erimen t design The experiment w as carefully designed to specifically address the heightened relev ance of force control feedbac k in the later stages of skull base pro cedures, where exc essiv e force can ha ve critical implications. The initial mastoidectomy w as conducted by tw o attending surgeons to derive expected forces during drilling around v arious anatom- ical structures, including cortical b one, trab ecular b one, and critical structures such 8 T able 1 Prop osed control related and anatomical structure related parameters σ high σ contact σ low η 1.7 0.7 0.3 1.0 index name γ n [mm] λ n [N] 1 F acial Nerve 1.5 0.8 2 T egemen 1.5 0.8 3 Sigmoid 1.5 0.8 4 Cortical 0.0 1.3 5 T rabecul ar 0.0 1.3 as the tegmen, sigmoid, and facial nerve. The statistical v alues obtained from drilling w ere then used to establish safet y threshold v alues for the force control scheme, with the statistical mean c hosen as the desired force. T o ensure a desired safet y margin for eac h individual anatomical structure and afford the rob ot sufficient time to preven t in teraction forces from reaching safety limits, the control metho ds w ere activ ated 0.2 N b efore the safet y margin limit. This activ ation threshold served as the trigger for initiating the adaptiv e drill force control, as outlined in T able 1 . The gain settings for in teraction force concerning different b one t yp es were determined through a prelimi- nary experiment, as detailed in T able 1 . Subsequently , the task of further skeletonizing the anatomical structures was entrusted to one medical student and tw o engineering studen ts with no prior exp erience in surgical drilling. 4.3 Result and discussion Throughout the exp erimen t, no c hattering or jittering was observ ed during the drilling (Fig. 4 ). During assessment of the exp erimen t, we fo cused on four key metrics: statistics related to undesired forces( F T > λ n + 0 . 2[N] ), statistics related to high forces( F T > λ n [N]), statistics of the av erage in teraction force ( F T > C = 0 . 3[N]), and the duration of time sp en t ab o ve the safety limit. The results of the prop osed control metho d are listed in Fig. 5 . The results sho w that the prop osed method can effectively limit the in teraction forces acting ab o ve the predefined safety limits for P1 and P2. F urthermore, it significan tly reduced the prop ortion of time sp en t ab ov e the estab- lished safety limit, considering the total duration of drilling. How ever, for P3, while Fig. 4 P ositional data (x, y , z) of the drill from a participan t’s procedure, with a green line indicating drill contact with anatomy (High signifies contact) and lab eled op erating structures. 9 T able 2 T abular results for the av erage proportion of time sp ent ab o ve the safety limit. Assist F acial Nerve T egmen Sigmoid Cortical T rab ecular w/o 0.726 0.549 0.567 0.372 0.209 w 0.322 0.370 0.382 0.243 0.042 Fig. 5 Comparison of the interaction force with and without prop osed metho d. Contact, High and Undesired denote when the tissue interaction force being ( F T > C = 0 . 3[N]), ( F T > λ n [N]) and ( F T > λ n + 0 . 2[N]) resp ectiv ely . Lower value indicates b etter result. op erating around the T egmen, Sigmoid, and Corticol, there was an increase in inter- action forces. This increase can b e attributed to p oor hand-ey e co ordination while p erforming the surgical task of skeletonizing the structure, as P3 w as observ ed to mak e sudden, fast in teractions with the anatomy during the exp erimen t. Nonetheless, these findings highlight a limitation of our propos ed metho d when users mak e sudden unin- tended in teractions with the anatomy , emphasizing the imp ortance of users possessing basic kno wledge of the surgical task. It is worth noting that these sudden, undesired in teractions can b e regulated b y employing virtual fixtures around critical anatomical structures [ 23 ], suggesting a p oten tial av enue for addressing suc h challenges in future implemen tations. 5 Conclusion In this study , w e ha v e in tro duced an adaptive force control metho d designed to activ ely limit interaction forces within the c hallenging surgical en vironment of skull base pro- cedures. Our prop osed situational-a ware force con trol emp o wers the rob ot to execute constrained motions in the presence of elev ated interaction forces while allowing the user to operate freely when forces are within safe op erating ranges. Additionally , it enhances the sensation of touc h when making contact with anatomical structures. This adaptabilit y is achiev ed through the automatic adjustment of admittance gains based on real-time assessmen ts of contextual information within the DT environmen t. The results from our initial exp erimen ts, inv olving a medical student and t wo engineering studen ts w orking with a cadav eric temp oral b one, clearly demonstrate the effectiveness of our prop osed metho d in successfully reducing undesired to ol-tissue in teractions. Our future research will encompass several key areas: (1) Threshold Refinement: While our initial study provided threshold v alues, we recognize the need for a more 10 extensiv e inv estigation in volving a larger sample to refine these thresholds. This will ensure that our system’s safety limits are w ell-calibrated to meet the diverse demands of clinical practice; (2) In tegration of Safety-Driv en Virtual Fixtures: W e will explore the in tegration of safet y-driven virtual fixtures aimed at aiding users in av oiding critical structures. This additional lay er of safet y guidance will further enhance the precision and safety of rob otic-assisted surgical pro cedures; (3) Comprehensive User Studies: W e plan to rigorously ev aluate the system’s p erformance through extensiv e user studies, inv olving participan ts with v arying lev els of surgical exp erience. W e aim to comprehensively assess how this technique con tributes to the o verall surgical exp erience, particularly in terms of safety and precision. Supplemen tary information. A supplementary video is pro vided with the sub- mission. Ac knowledgmen ts. Nimesh Nagururu is supp orted in part by NCA TS TL1 Grant TR003100. This work w as also supp orted in part by a research con tract from Galen Rob otics, by NIDCD K08 Grant DC019708, by a research agreemen t with the Hong Kong Multi-Scale Medical Rob otics Centre, and b y Johns Hopkins Universit y internal funds. Declarations Under a license agreement b et ween Galen Rob otics, Inc and the Johns Hopkins Univ ersity , Russell H. T aylor and Johns Hopkins Universit y are en titled to ro yalt y distributions on tec hnology that may p ossibly be related to that discussed in this pub- lication. Dr. T aylor also is a paid consultant to and owns equit y in Galen Rob otics, Inc. This arrangement has been review ed and approv ed by Johns Hopkins Universit y in accordance with its conflict-of-interest p olicies. References [1] Meyb o di, A.T., Mignucci-Jim ´ enez, G., La wton, M.T., Liu, J.K., Preul, M.C., Sun, H.: Comprehensive microsurgical anatom y of the middle cranial fossa: P art I—Osseous and meningeal anatomy . F ron tiers in Surgery 10 (2023) [2] Jain, S., Deshmukh, P ., Lakhotia, P ., Kalambe, S., Chandrav anshi, D., Khatri, M.: Anatomical study of the facial recess with implications in round window visibil- it y for co c hlear implantation: personal observ ations and review of the literature. In ternational Archiv es of Otorhinolaryngology 23 , 281–291 (2019) [3] Kratchman, L.B., Sch uster, D., Dietric h, M.S., Labadie, R.F.: F orce p erception thresholds in co c hlear implan tation surgery . Audiology and Neurotology 21 (4), 244–249 (2016) [4] Gantz, B.J.: Ev olution of otology and neurotology education in the United States. Otology & Neurotology 39 (4S), 64–68 (2018) 11 [5] T rejos, A., Patel, R., Naish, M.: F orce sensing and its application in minimally in v asive surgery and therapy: a survey . Journal of Mechanical Engineering Science 224 (7), 1435–1454 (2010) [6] Simaan, N., T aylor, R.H., Choset, H.: Intelligen t surgical rob ots with situational a wareness. Mec hanical Engineering 137 (09), 3–6 (2015) [7] Attanasio, A., Scaglioni, B., De Momi, E., Fiorini, P ., V aldastri, P .: Autonomy in surgical rob otics. Ann ual Review of Control, Robotics, and Autonomous Systems 4 , 651–679 (2021) [8] Golahmadi, A.K., Khan, D.Z., Mylonas, G.P ., Marcus, H.J.: T o ol-tissue forces in surgery: A systematic review. Annals of Medicine and Surgery 65 , 102268 (2021) [9] Haidegger, T., Ben y´ o, B., Kov´ acs, L., Beny´ o, Z.: F orce sensing and force control for surgical rob ots. IF A C Pro ceedings V olumes 42 (12), 401–406 (2009) [10] Suomalainen, M., Karayiannidis, Y., Kyrki, V.: A surv ey of robot manipulation in contact. Rob otics and Autonomous Systems 156 , 104224 (2022) [11] Minniti, M.V., Grandia, R., F¨ ah, K., F arshidian, F., Hutter, M.: Mo del predictiv e rob ot-en vironmen t interaction control for mobile manipulation tasks. In: ICRA, pp. 1651–1657 (2021). IEEE [12] Y en, P .-L., Chen, Y.-J.: Con tact compliance based visual feedback for to ol alignmen t in robot assisted bone drilling. Sensors 22 (9), 3205 (2022) [13] Hu, Y., Jin, H., Zhang, L., Zhang, P ., Zhang, J.: State recognition of p edi- cle drilling with force sensing in a rob otic spinal surgical system. IEEE/ASME T ransactions on Mec hatronics 19 (1), 357–365 (2014) [14] Li, H., Nie, X., Duan, D., Li, Y., Zhang, J., Zhou, M., Magid, E.: An admittance- con trolled amplified force trac king scheme for collab orativ e lumbar puncture surgical rob ot system. IJMR CAS 18 (5), 2428 (2022) [15] Xie, Y., Sun, D., Liu, C., Tse, H.Y., Cheng, S.H.: A force con trol approac h to a rob ot-assisted cell microinjection system. IJRR 29 (9), 1222–1232 (2010) [16] Rothbaum, D.L., Ro y , J., Stoianovici, D., Berkelman, P ., Hager, G.D., T aylor, R.H., Whitcomb, L.L., F rancis, H.W., Niparko, J.K.: Rob ot-assisted stapedo- tom y: micropic k fenestration of the stap es fo otplate. Otolaryngology-head and nec k surgery 127 (5), 417–426 (2002) [17] Rothbaum, D.L., Roy , J., Hager, G.D., T aylor, R.H., Whitcom b, L.L., F rancis, H.W., Niparko, J.K.: T ask performance in stap edotomy: Comparison b et ween surgeons of differen t exp erience levels. Otolaryngology—Head and Neck Surgery 128 (1), 71–77 (2003) 12 [18] Ebrahimi, A., P atel, N., He, C., Gehlbach, P ., Kobilaro v, M., Iordachita, I.: Adap- tiv e con trol of sclera force and insertion depth for safe rob ot-assisted retinal surgery . In: ICRA, pp. 9073–9079 (2019). IEEE [19] Sang, H., Monfaredi, R., Wilson, E., F o oladi, H., Preciado, D., Cleary , K.: A new surgical drill instrument with force sensing and force feedbac k for rob otically assisted otologic surgery . Journal of Medical Devices 11 (3) (2017) [20] Maier, T., Strauss, G., Hofer, M., Kraus, T., Runge, A., Stenzel, R., Gumprech t, J., Berger, T., Dietz, A., Lueth, T.C.: A new micromanipulator system for middle ear surgery . In: 2010 IEEE In ternational Conference on Rob otics and Automation, pp. 1568–1573 (2010). IEEE [21] Miroir, M., Nguyen, Y., Szewczyk, J., Mazalaigue, S., F errary , E., Sterkers, O., Gra yeli, A.B.: Rob otol: from design to ev aluation of a rob ot for middle ear surgery . In: 2010 IEEE/RSJ International Conference on Intelligen t Robots and Systems, pp. 850–856 (2010). IEEE [22] Chen*, Y., Go odridge*, A., Sahu*, M., Kishore, A., V afaee, S., Mohan, H., Sap ozhnik o v, K., Creighton, F.X., T aylor, R.H., Galaiya, D.: A force-sensing sur- gical drill for real-time force feedback in rob otic mastoidectom y . IJCARS, 1–8 (2023) [23] Ishida*, H., Sahu*, M., Muna war, A., Nagururu, N., Galiya, D., Kazanzides, P ., Creigh ton, F.X., T aylor, R.H.: Haptic-Assisted Collab orativ e Rob ot F ramework for Impro ved Situational Awareness in Skull Base Surgery . Under Review for ICRA2024. [24] Shu*, H., Liang*, R., Li*, Z., Go odridge, A., Zhang, X., Ding, H., Nagururu, N., Sah u, M., Creigh ton, F.X., T aylor, R.H., et al.: Twin-s: a digital twin for skull base surgery . IJCARS, 1–8 (2023) [25] Olds, K.C., Chalasani, P ., Pac heco-Lop ez, P ., Iordachita, I., Akst, L.M., T aylor, R.H.: Preliminary ev aluation of a new microsurgical rob otic system for head and nec k surgery . In: ICRA, pp. 1276–1281 (2014). IEEE [26] Olds, K.: Rob otic assistant systems for otolaryngology-head and nec k surgery . PhD thesis, Johns Hopkins Universit y (2015) [27] Keemink, A.Q., Ko oij, H., Stienen, A.H.: Admittance control for ph ysical h uman– rob ot in teraction. IJRR 37 (11), 1421–1444 (2018) [28] Ishida*, H., Barragan*, J.A., Munaw ar, A., Li, Z., Ding, A., Kazanzides, P ., T raki- mas, D., Creighton, F.X., T a ylor, R.H.: Impro ving surgical situational a wareness with signed distance field: a pilot study in virtual reality . In: IR OS, pp. 8474–8479 (2023). IEEE 13