Stable Tracking-in-the-Loop Control of Cable-Driven Surgical Manipulators under Erroneous Kinematic Chains

Neelay Joglekar, Fei Liu, Florian Richter, Michael C. Yip · 2025

Read Stable Tracking-in-the-Loop Control of Cable-Driven Surgical Manipulators under Erroneous Kinematic Chains by Neelay Joglekar, Fei Liu, Florian Richter, Michael C. Yip (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-05663.txt.

Introduction

Surgical robots driven by cables can place tools with great care inside the body, but cable stretch and joint misreadings corrupt the math that steers them. Neelay Joglekar, Fei Liu, Florian Richter, and Michael Yip addressed a blind spot in 2025. Joints before the insertion point cannot be seen by the camera, so their errors cannot be corrected by vision.

The team designed a controller that keeps tracking stable even with those unseen errors, and proved its stability. They folded it into a two-level scheme for the whole arm. Tests in simulation and on real hardware matched the theory. The work supports future autonomous subtasks such as suturing and suction.

Catalog record

CREATORNeelay Joglekar, Fei Liu, Florian Richter, Michael C. Yip
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-05663

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Stable Tracking-in-the-Loop Control of Cable-Driven Surgical Manipulators under Erroneous Kinematic Chains Neelay Joglekar, Fei Liu, Florian Richter, Michael C. Yip Abstract Remote Center of Motion (RCM) robotic manipulators have revolutionized Minimally Invasive Surgery, enabling precise, dexterous surgical manipulation within the patient’s body cavity without disturbing the insertion point on the patient. Accurate RCM tool control is vital for incorporating autonomous subtasks like suturing, blood suction, and tumor resection into robotic surgical procedures, reducing surgeon fatigue and improving patient outcomes. However, these cable-driven systems are subject to significant joint reading errors, corrupting the kinematics computation necessary to perform control. Although visual tracking with endoscopic cameras can correct errors on in-view joints, errors in the kinematic chain prior to the insertion point are irreparable because they remain out of view. No prior work has characterized the stability of control under these conditions. We fill this gap by designing a provably stable tracking-in-the-loop controller for the out-of-view portion of the RCM manipulator kinematic chain. We additionally incorporate this controller into a bilevel control scheme for the full kinematic chain. We rigorously benchmark our method in simulated and real world settings to verify our theoretical findings. Our work provides key insights into the next steps required for the transition from teleoperated to autonomous surgery. I INTRODUCTION Robotic surgical systems provide surgeons more precise, tremor-free control of surgical tools, enabling numerous Robotically Assisted Minimally Invasive Surgery (RAMIS) procedures that improve patient outcomes. This is particularly due to their use of Remote Center of Motion (RCM) manipulators [ 1 ] , which can freely move about the surgical workspace while remaining stationary at an insertion point. Although current RAMIS procedures are fully teleoperated, automating surgical subtasks such as suturing [ 2 , 3 ] , tumor resection [ 4 , 5 ] , tissue grasping [ 6 ] , and blood suction [ 7 , 8 , 9 ] can reduce surgeon workload and enable them to focus on higher-level decision making. Many prior works have proposed and demonstrated autonomous solutions for these subtasks, but a key roadblock to their implementation has been lack of confidence in their reliability. As a result, there has been growing interest in developing reliability-driven autonomous algorithms [ 10 , 11 , 12 ] that can confidently handle any uncertainty or risk that may arise in surgical situations. Fig. 1: Cable-Driven Remote Center of Motion (RCM) surgical manipulators are prone to kinematic chain errors, and these cannot be fully corrected because a large portion of the robot remains out of view of the endoscopic camera. This can produce unstable control behavior. We are the first to provide theoretical analysis on control stability under errors in the RCM kinematic chain, formulating an associated controller with strong stability guarantees. (Torso image altered from [ 13 ] with GPT-4o and DALLE to remove annotations). Enforcing reliable manipulator control is a key prerequisite in ensuring the safety of any autonomous procedure, but this is not straightforward for serial RCM surgical systems. A typical kinematics-based closed-loop controller assumes access to accurate ground-truth joint angle readings. However, in cable-driven RCM manipulators, joint angle measurements are often erroneous due to cable effects such as hysteresis, friction, and backlash. Adding visual tracking into the loop can help correct joint reading errors in observable joints, but several RCM manipulator joints remain out-of-view, as shown in Fig. 1 , and cannot be individually corrected [ 14 ] . This produces significant error in the kinematic chain, as shown in Fig. 2 . How does this untracked kinematic error contribute to errors in trajectory following and, under visual closed-loop control, can it produce an unstable control system? No prior work has sought to answer this question, or provide theoretical guarantees on control stability under these conditions, jeapordizing the reliability of RCM manipulator control algorithms. In this manuscript, we seek to fill this gap, providing the following contributions: • The first theoretical analysis of tracking-in-the-loop RCM manipulator control stability under kinematic chain inaccuracy; • A resolved-rate controller for the out-of-view component of the RCM kinematic chain that is theoretically stable within a wide joint error bound; • A bilevel control scheme that incorporates our stable out-of-view chain controller into a full chain control algorithm. We validate our results in simulation and hardware experiments, using the da Vinci Research Kit (dVRK) as a case study. Our work provides crucial insight into the next steps required to the improve the reliability of surgical autonomy. II RELATED WORKS Initial works used sophisticated calibration techniques to account for cable effects in RCM surgical tool control [ 15 , 16 ] . For example, Hwang et al. used fiducial attachments to collect a large dataset of tool poses and joint angles, which was used to train an RNN-based controller. However, calibrations degrade over time through cable creep, where cables irreversibly stretch over time. To address this, other works incorporate visual tracking into the loop. Although it would be ideal to track camera extrinsic calibration error and each joint angle error as a separate term, Richter et al. proved it’s typically impossible to estimate these quantities individually [ 14 ] . Instead, most works track the combined effect of these errors [ 17 , 18 ] , often referred to either as the “kinematics remote-center coordinate system (KCS)” [ 19 , 20 ] or the “lumped” error correction term [ 14 , 21 ] . Although this approach enables accurate tool tip tracking, it produces a false estimate of the full kinematic chain, as shown in Fig. 2 . Prior works naively relied on this false chain to perform control without providing guarantees on stability [ 7 , 22 , 23 ] . As will be shown in Section V, it is indeed possible for visual-servoing based controllers to become unstable when these kinematic chain errors are not properly handled. We seek to fill this knowledge gap in this manuscript, providing insight into how lumped error tracking alters the estimated kinematic chain and performing theoretical analysis on the stability of controlling this chain. III BACKGROUND Fig. 2: A 2D example: a) An ideal kinematic chain can be constructed from joint readings and a base-to-camera transform. b) However, camera calibration and joint reading errors produce an erroneous kinematic chain (red). c) “Lumped error” tracking paired with visual joint angle correction repairs the configuration of the in-view chain (green), but the out-of-view chain (orange) remains inaccurate. This kinematic inaccuracy complicates control, especially since the “lumped error” term (orange dashed) dynamically changes with the joint angles. III-A Lumped Error Tracking Visualized in Fig. 2 (a), the kinematic chain for an ideal manipulator is as follows. \mathbf{T}=\mathbf{T}^{c}_{b}\prod_{i=1}^{n}\mathbf{T}^{i-1}_{i}(q_{i}) (1) where \mathbf{q}=[q_{1}\dots q_{n}]^{\top} are the true joint angles, \mathbf{T}^{i-1}_{i}(q_{i}) is the transform from link i to i-1 , and \mathbf{T}^{c}_{b} is the base-to-camera transform. However, as shown in Fig. 2 (b), surgical manipulators typically deal with two types of errors: errors in the base-to-camera calibration and errors in the joint angle readings. In this case, the true chain can be rewritten as follows: \mathbf{T}=\mathbf{T}^{c}_{b-}\mathbf{T}^{b-}_{b}\prod_{i=1}^{n}\mathbf{T}^{i-1}_{i}(\tilde{q}_{i}+e_{i}) (2) Here, \mathbf{\tilde{q}}=[\tilde{q}_{1}\dots\tilde{q}_{n}]^{\top} are joint angle readings with errors \mathbf{e}=[e_{1}\dots e_{n}]^{\top} , such that \mathbf{q}=\mathbf{\tilde{q}}+\mathbf{e} , and \mathbf{T}^{c}_{b-} is the base-to-camera transform combined with the camera extrinsic calibration error \mathbf{T}^{b-}_{b} .

Read the complete text