Pith. sign in

REVIEW 3 major objections 4 minor 59 references

This paper claims that a combined reactive and iterative-learning controller can control redundantly actuated multilink and hybrid cable-driven parallel robots in operational space in real time, resolving both kinematic and actuation redund

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 05:45 UTC pith:QNLA4URP

load-bearing objection A practical tri-space control framework for cable-driven parallel robots that works in simulation and hardware, but the convergence theorem rests on an assumption that cannot hold for any feasible trajectory. the 3 major comments →

arxiv 2607.29500 v1 pith:QNLA4URP submitted 2026-07-31 cs.RO cs.SYeess.SY

Tri-Space Operational Control of Redundant Multilink and Hybrid Cable-Driven Parallel Robots Using an Iterative-Learning based Reactive Approach

Dipankar Bhattacharya , Yin Pok Chan , Siqi Shang , Yuen Shan Chan , Ying Tan , Darwin Lau This is my paper
classification cs.RO cs.SYeess.SY
keywords cable-driven parallel robotstri-space controlreactive controliterative learning controlkinematic redundancyactuation redundancynull space parameterizationcable-link interference
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes a two-timescale control scheme for cable-driven robots that have more actuators than joints and more joints than task coordinates. A fast reactive controller solves a quadratic program at each step to track a desired end-effector path while keeping cable forces positive and steering clear of cable-link collisions, singular configurations, and joint limits. A slower iterative-learning layer tunes a small set of parameters — the null-space scaling of the joint acceleration command and the weights of the cost terms — across repeated executions of the same task, improving tracking and cutting actuation effort. The paper argues this is the first tri-space framework that works online for several architectures of multilink and hybrid cable-driven robots, and supports the claim with simulations on three different robots and hardware experiments on a two-link arm.

Core claim

For a cable-driven robot with two levels of redundancy, the paper's claim is that both the kinematic redundancy (joint trajectories that achieve the same task motion) and the actuation redundancy (cable/joint force sets that achieve the same joint motion) can be resolved in one online quadratic program, with the null-space component of the joint acceleration parameterized as (I − J†_W J) D J†_W b(Xd, q̈), where D is a learned diagonal scaling matrix. The performance over repeated trials is then improved by learning the reactive tuning parameter θ = [d_1,…,d_n, ln α, ln β] that minimizes a weighted sum of tracking error and actuation effort. The framework is claimed to be directly applicable

What carries the argument

The load-bearing object is the null-space parameterization of the joint acceleration command: q̈_cN = (I − J†_W J) D J†_W b(Xd, q̈), in which an unknown diagonal matrix D scales the components of the nominal acceleration before projection into the null space of the joint–task Jacobian. This D, together with the cost weights α and β, forms an (n+2)-dimensional parameter vector θ that the iterative-learning controller updates once per trial using pattern search or particle swarm optimization. The quadratic-program reactive controller uses the parameterized command to solve for joint acceleration and actuation commands at every sample instant, with linear hard constraints and a soft avoidance f

Load-bearing premise

The convergence proof relies on Assumption 1: there is a known compact set of parameters containing a unique θ* at which the trajectory performance function P(θ*) is exactly zero — meaning perfect tracking with zero cable and joint actuation effort — which no non-constant trajectory can actually achieve.

What would settle it

Run the iterative-learning search over a dense grid of θ on a non-constant trajectory and compute the global minimum of P(θ); if that minimum is bounded away from zero, Assumption 1 is violated and Theorem 2's convergence conclusion does not apply. Alternatively, drive the hardware for many iterations and check whether the tracking error and actuation norms continue to decrease toward zero or plateau at a strictly positive floor.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • A single RC/ILC framework can be deployed on different cable-driven robot architectures without re-deriving the controller, since the QP and the parameterization are formulated with the generalized model.
  • Because ILC searches only n+2 parameters rather than the full actuation trajectory, the improvement over repetitions is computationally cheap enough to run online between trials.
  • The avoidance functions let the robot detect and steer away from low manipulability, cable-link interference, and joint limits before they become hard constraint violations.
  • On the BMArm hardware, the learned parameters reduced the performance function by 64% and the maximum tracking error norm from 0.014 m to 0.006 m.
  • The framework converts constraint avoidance into learned joint-space behavior: over iterations it discovers trajectories that stay outside the avoidance transition region altogether, improving both error and actuation effort.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The low-dimensional parameterization suggests the same scheme could be applied to other redundantly actuated mechanisms (tendon-driven hands, musculoskeletal arms) where the null space is large and full-trajectory learning is infeasible.
  • One could test whether the learned null-space parameters transfer from one trajectory to a neighbouring one on the same robot, which would indicate the parameters capture something about the robot's geometry rather than the specific path.
  • The formal convergence guarantee assumes a parameter vector that achieves zero tracking error and zero actuation effort; a mathematically grounded refinement would replace that with a known positive lower bound and prove convergence to that bound.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes a unified reactive + iterative-learning control framework for redundant multilink cable-driven parallel robots (MCDRs) and hybrid cable-driven robots (HCDRs). The reactive layer solves a convex QP at each sampling instant to determine joint accelerations and actuation commands that track an operational-space trajectory while enforcing cable-force/torque bounds and avoidance constraints (manipulability, cable-link interference, joint limits). The ILC layer tunes a low-dimensional parameter vector θ (null-space scaling scalars d_i and the cost weights α, β) between trajectory repetitions using pattern search or PSO, minimizing a scalar performance index that combines normalized tracking error and actuation effort. The framework is tested in simulation on the BMArm, SpiderArm, and FASTKIT-Planar robots and in hardware on the BMArm, with quantitative before/after ILC comparisons. The central claim is that this is the first general tri-space control framework that handles both kinematic and actuation redundancy, avoids undesirable situations in real time, and improves performance over repeated executions.

Significance. If the practical claims hold, the framework is a potentially valuable contribution: it provides a unified treatment of two-level redundancy for several CDPR architectures, with open-source CASPR implementation, hardware validation, and a low-dimensional ILC parameterization that avoids learning full actuation trajectories. The empirical demonstrations show meaningful performance improvements. However, the formal stability and convergence analysis is not reliable as stated: the main convergence theorem relies on an assumption that is incompatible with the system's own actuation constraints, and the QP feasibility that underpins the boundedness theorem is only assumed, not established. The practical methodology is defensible, but the manuscript's theoretical claims overstate what is proven.

major comments (3)
  1. [Section VI (Assumption 1) and Section VII-B (Theorem 2)] Assumption 1 asserts the existence of a unique θ* with P(θ*)=0. But the trajectory performance function P in (40)-(43) includes the cable actuation effort term P_C(θ)=||A_C(θ)||_F. Since cable forces are constrained by (4) and (13) to satisfy f >= f_min > 0, for any trajectory with at least one sample P_C(θ) >= sqrt(N) f_min > 0 for every θ. Therefore P(θ) > 0 for all θ, and Assumption 1 cannot hold for any of the systems/experiments considered in Section VIII. Theorem 2, which concludes limsup_{i→∞} ||e_i(t)|| <= δ for arbitrary δ>0, is derived from this false premise, so the stated convergence guarantee is vacuous. Remark 7 already concedes that only boundedness follows when P(θ*) != 0. The authors should either remove Theorem 2 and state a boundedness/ultimate-boundedness result, or give a nontrivial condition under which P(θ*) = 0 can be satisfied (which appears impossible under posi
  2. [Section V-A (Eq. (25)) and Section VII-A (Theorem 1)] The stability analysis assumes that the QP (25) is feasible at every sampling instant and that Algorithms 1-2 produce a_i^k and ¨q_i^k. No conditions are given that guarantee feasibility of the combined constraints (cable force bounds, joint torque bounds, and the hard avoidance constraints (28)-(30)). The paper's own simulations show that the reactive controller can fail or diverge for certain initial conditions or trajectories (e.g., Row 2 of Table I for BMArm; FASTKIT-Planar rows in Table II where the 'before ILC' controller fails). Thus the boundedness result in Theorem 1 and the convergence argument in Theorem 2 are conditional on an unverified assumption that is load-bearing for the claim that the framework 'ensures feasible solutions'. This should be stated explicitly as an assumption with a discussion of when it can be satisfied, or proven under restricted admissible sets.
  3. [Section IV-C, Eq. (22)] The claimed null-space parameterization is not complete. For a fixed vector b, the set {(I−J†_W J) D J†_W b : D diagonal} is only an n-dimensional linear family within the (n−r)-dimensional null space of J; multiplying the particular solution by a diagonal matrix and projecting does not, in general, generate all possible null-space vectors. Consequently, the θ* found by ILC is optimal only within this restricted parameterization, not globally over all kinematically feasible joint trajectories. This should be qualified in the text; otherwise the statement 'finding an optimal parameter in the null space' is misleading.
minor comments (4)
  1. [Section VI-A, normalization] The normalization uses P*_E = min_{θ∈Θ} P_E(θ) (and similarly for P_C, P_D), with the minimum taken over parameters computed so far. If the minimum is zero (e.g., P_E can vanish under perfect tracking), the normalized terms are undefined. Please specify safeguards for this case, such as a small positive floor.
  2. [Throughout] There are several typographical errors: 'intergern' in Section V-A, 'quarternions' in Section III-B, and duplicated figure caption text in Fig. 8. These should be corrected.
  3. [Table II] The FASTKIT-Planar star and cylinder-sine-wave rows have formatting irregularities (e.g., missing entries and non-aligned columns). Please verify the table layout and ensure all performance values are reported consistently.
  4. [Sections VIII-IX] All quantitative comparisons are based on single simulation or hardware runs. Given that sensor noise is added stochastically in simulation and that the ILC uses stochastic PSO in hardware, reporting mean and standard deviation over several runs would strengthen the robustness claims.

Circularity Check

1 steps flagged

Formal convergence theorem is self-definitional via the definition of P; the practical RC/ILC framework itself is not circular.

specific steps
  1. self definitional [Section VI Assumption 1; Section VI-A Eqs. (40)-(43); Section VII-B Theorem 2 and Remark 7]
    "Assumption 1. If the solution of the QP formulation (25) is feasible for any given initial θ0, there exists a compact set ΘC such that the optimal θ∗ exists uniquely in this compact set ΘC satisfying P(θ∗) = 0. ... PE(θ) = ∥E(θ)∥F ... P(θ) = ρe P̂E(θ) + ρc P̂C(θ) + ρd P̂D(θ) ... If Assumption 1 holds, Algorithm 1 can ensure that lim i→∞ P(θ∗ i) = 0 and lim i→∞ θ∗ i = θ∗, indicating that lim i→∞ E(θ∗ i) = 0, which comes from (41) and (43)."

    P(θ) is defined in (40)-(43) as a positive-weighted sum of the normalized tracking-error norm, cable-effort norm, and direct-effort norm. Hence P(θ∗)=0 already asserts E(θ∗)=0, i.e., zero tracking error at every sample. Theorem 2's conclusion limsup ||e_i(t)|| ≤ δ is then the assumption restated together with Algorithm 1's built-in monotone-decrease acceptance rule, not an independent guarantee. Moreover, cable forces are constrained positive by (4)/(13), so P_C(θ) ≥ √N f_min > 0 for every feasible θ, making Assumption 1 unsatisfiable for the paper's own systems. Remark 7 concedes that when P(θ∗) ≠ 0 only boundedness follows, so the formal convergence claim adds no content beyond the definition of P.

full rationale

The central practical architecture is not circular. The ILC explicitly solves θ∗ = argmin P(θ), so the reported reduction in P and tracking error over iterations is the intended learning behavior, not a hidden fit or a prediction of unseen data. The null-space parameterization (22)-(23), the RC quadratic program (25), and the avoidance formulations (28)-(37) are free design choices with independent content. The simulations use the open-source CASPR platform and the BMArm hardware experiments provide external validation; self-citations to CASPR [30], generalized modeling [42], and inverse dynamics [38] are infrastructure/model references rather than load-bearing uniqueness claims, and they do not smuggle in the framework's conclusions. The genuine circular element is the formal convergence guarantee: Assumption 1 postulates P(θ∗)=0, but by (40)-(43) P is defined from the tracking-error norm E and actuation-effort norms, so P=0 already contains E=0, and the proof then derives E→0 'which comes from (41) and (43)'. Since cable forces are positive by (4)/(13), P is actually >0 for every feasible θ, so the premise is unsatisfiable; Remark 7 admits only boundedness in that case. Thus Theorem 2's guarantee is either definitionally contained in its own premise or vacuous. This is a real weakness in the formal analysis, but it does not reduce the overall empirical framework to a fit, so the circularity is partial rather than total.

Axiom & Free-Parameter Ledger

10 free parameters · 6 axioms · 0 invented entities

The central claim depends on user-chosen and learned parameters (θ, α, β, PD gains, avoidance thresholds), a known dynamic model, QP feasibility at every step, and an unrealistic uniqueness/zero-cost assumption for the convergence theorem. The paper's contribution is the framework and its empirical demonstrations, not a parameter-free derivation.

free parameters (10)
  • Null-space scaling matrix D (parameters d_1..d_n in θ) = Learned; hardware bounds [-2,2] per d_i; simulations report u2 increased and u4 decreased, no final values tabulated
    Parameterizes q¨cN in (22)-(23); ILC/PS/PSO optimizes these to minimize P(θ); central to exploiting kinematic redundancy.
  • Cost weights α and β (θ entries n+1, n+2 via log) = Initialized 1e-6 and 0/0.01/0.1/1 across experiments; learned by ILC
    Balance tracking vs actuation vs avoidance in V (21); included in learned θ (24); affect all reported results.
  • PD gains Kp, Kd = Table I: e.g., 200I3, 28I3; hardware 20000I3, 282I3
    Define the ideal closed-loop (16)-(18); chosen by hand; authors state choice affects transients but not stability (Remark 1).
  • Avoidance acceleration gain k1 = Not reported
    Sets desired velocity magnitude in (32); governs how strongly the robot avoids undesirable configurations; no tuning study provided.
  • Cable-link interference buffer ε and hard limit = ε=0.2 m, hard limit 0.1 m for SpiderArm
    Defines avoidance transition region in (37) and hard constraints (28)-(30); directly affects whether interference is avoided.
  • Weight function steepness λ_w = Not reported
    Controls w(q) in (37); determines how abruptly cable-link avoidance is activated.
  • Performance normalization weights ρ_e, ρ_c, ρ_d = Set to 1 in simulations
    Form P(θ) in (43); user-chosen; different choices would change what ILC optimizes.
  • Pattern search step size δθ and convergence rate ρ_ILC = Step size 1 initially; ρ_ILC not tabulated
    Algorithm 1 parameters govern ILC search and the monotone decrease condition.
  • Compact search set Θ and PSO settings = Hardware θ_min=[-2,-2,-2,-2,ln(1e-7),ln(0.1)], θ_max=[2,2,2,2,ln(1e-5),ln(10)]; PSO 6 particles, ω=0.73, φ_p=φ_g=1.5
    Bounds and optimizer settings constrain learned θ; hardware results depend on this domain knowledge.
  • Actuation weighting matrix W_a = Not specified exactly; positive definite, typically diagonal
    Appears in g_f(a)=a^T W_a a (27); user-chosen trade-off on which actuators are penalized.
axioms (6)
  • domain assumption CDPR dynamic model (11) with known M, C, G and B is exact for the robots.
    Used as equality constraint in QP (25) and in stability analysis; unmodeled friction/backlash are only partially compensated by high PD gains in hardware.
  • domain assumption The reactive QP (25) is feasible at every sampling instant and every iteration for the initial θ0.
    Algorithm 2 requires a QP solution at each k; no feasibility proof is given, and failure cases without avoidance demonstrate infeasibility can occur (Section VIII-A).
  • ad hoc to paper Assumption 1: an optimal θ* exists uniquely in a known compact set Θ_C with P(θ*)=0.
    Used to prove Theorem 2; because P includes tracking error and actuation effort norms, P=0 is unattainable for the tested non-constant trajectories.
  • domain assumption The repeated task is iteration-invariant and iteration-varying noise is not dominant (Remark 4).
    ILC assumes the optimal θ* is fixed across identical repetitions; hardware friction and noise violate this to some degree, motivating the choice of PSO.
  • domain assumption The ideal PD closed-loop (17) is attainable, so b(·) in (18) corresponds to desired operational-space behavior.
    RC minimizes distance to J†_W b; if constraints bind, the PD ideal is not realized and stability relies on boundedness of QP solutions.
  • standard math Sampled-data stabilization theorem [57, Thm 1] applies under the controller's zero-order-hold actuation (44).
    Theorem 2's error bound depends on this external result; authors do not verify its conditions beyond asserting boundedness.

pith-pipeline@v1.3.0-daily-deepseek · 30025 in / 15440 out tokens · 153255 ms · 2026-08-03T05:45:34.430997+00:00 · methodology

0 comments
read the original abstract

Cable-Driven Parallel Robots (CDPRs) are a type of parallel mechanism in which cables are used as actuators. Due to the two levels of redundancy and numerous constraints within the CDPR actuation, joint and operational spaces (together known as the tri-space), tracking a given trajectory in the operational space while satisfying constraints in tri-space simultaneously is challenging. To the best of the authors' knowledge, there does not exist any tri-space control framework, which is robust, effective, and directly applicable to several architectures of redundantly actuated CDPRs. This paper proposes a tri-space control framework that combines Reactive Control (RC) and Iterative-Learning Control (ILC) to perform repetitive tasks in the operational space. The framework allows the tracking of operational space trajectories online with feasible cable forces, while avoiding undesirable situations such as cable-link interference, joint interference, and loss of manipulability. On the other hand, by finding an optimal parameter in the null space using a novel parameterization of a null space vector, the performance can be improved through ILC when the task is repeatedly executed. Simulation and hardware results on various Multilink Cable-Driven Robot (MCDRs) and Hybrid Cable-Driven Robots (HCDRs) show that the proposed tri-space control framework can be conveniently and effectively applied to the real-time control of different CDPRs.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

59 extracted references

  1. [1]

    The NIST robocrane,

    J. S. Albus, R. V . Bostelman, and N. Dagalakis, “The NIST robocrane,” J. Robot. Syst., vol. 10, no. 5, pp. 709–724, 1993

    1993

  2. [2]

    A dual-stage planar cable robot: Dynamic modelling and design of a robust controller with positive inputs,

    S.-R. Oh, K. Mankala, S. K. Agrawal, and J. S. Albus, “A dual-stage planar cable robot: Dynamic modelling and design of a robust controller with positive inputs,”J. Mech. Des., vol. 127, no. 4, pp. 612–620, 2005

    2005

  3. [3]

    Cable-suspended robotic contour crafting system,

    P. Bosscher, R. L. Williams II, L. S. Bryson, and D. Castro-Lacouture, “Cable-suspended robotic contour crafting system,”Autom. Constr., vol. 17, no. 1, pp. 45–55, 2007

    2007

  4. [4]

    CU- brick cable-driven robot for automated construction of complex brick structures: From simulation to hardware realisation,

    Y . Wu, H. H. Cheng, A. Fingrut, K. Crolla, Y . Yam, and D. Lau, “CU- brick cable-driven robot for automated construction of complex brick structures: From simulation to hardware realisation,” inProc. IEEE Int. Conf. Sim. Model. Program. Auton. Robot., pp. 166–173, 2018

    2018

  5. [5]

    Design of a cable-driven arm exoskeleton (CAREX) for neural rehabilitation,

    Y . Mao and S. K. Agrawal, “Design of a cable-driven arm exoskeleton (CAREX) for neural rehabilitation,”IEEE Trans. Robot., vol. 28, no. 4, pp. 922–931, 2012

    2012

  6. [6]

    Large-scale 3d printing with cable-driven parallel robots,

    J.-B. Izard, A. Dubor, P.-E. Herv´e, E. Cabay, D. Culla, M. Rodriguez, and M. Barrado, “Large-scale 3d printing with cable-driven parallel robots,” Constr. Robots., vol. 1, no. 1, pp. 69–76, 2017

    2017

  7. [7]

    On the task specific evaluation and optimisation of cable-driven manipulators,

    D. Lau, K. Bhalerao, D. Oetomo, and S. K. Halgamuge, “On the task specific evaluation and optimisation of cable-driven manipulators,” inAdvances in Reconfigurable Mechanisms and Robots I, J. S. Dai, M. Zoppi, and X. Kong, Eds., ch. 63, pp. 707–716. Springer London, 2012

    2012

  8. [8]

    On the ability of a cable- driven robot to generate a prescribed set of wrenches,

    S. Bouchard, C. Gosselin, and B. Moore, “On the ability of a cable- driven robot to generate a prescribed set of wrenches,”J. Mech. Robot., vol. 2, no. 1, pp. 011 010/1–10, 2010

    2010

  9. [9]

    Analysis of bounded cable tensions in cable-actuated parallel manipulators,

    M. Hassan and A. Khajepour, “Analysis of bounded cable tensions in cable-actuated parallel manipulators,”IEEE Trans. Robot., vol. 27, no. 5, pp. 891–900, 2011

    2011

  10. [10]

    Conditions on the cable-routing matrix for wrench closure of multilink cable-driven manipulators,

    D. Lau and D. Oetomo, “Conditions on the cable-routing matrix for wrench closure of multilink cable-driven manipulators,”J. Mech. Des., vol. 138, no. 3, pp. 032 303/1–7, 2016

    2016

  11. [11]

    Interval-analysis-based determination of the wrench-feasible workspace of parallel cable-driven robots,

    M. Gouttefarde, D. Daney, and J.-P. Merlet, “Interval-analysis-based determination of the wrench-feasible workspace of parallel cable-driven robots,”IEEE Trans. Robot., vol. 27, no. 1, pp. 1–13, 2011

    2011

  12. [12]

    Workspace analysis of multibody cable-driven mechanisms,

    S. Rezazadeh and S. Behzadipour, “Workspace analysis of multibody cable-driven mechanisms,”J. Mech. Robot., vol. 3, no. 2, pp. 021 005/1– 10, 2011

    2011

  13. [13]

    Cable suspended robots: Design, planning and control,

    A. B. Alp and S. K. Agrawal, “Cable suspended robots: Design, planning and control,” inProc. IEEE Int. Conf. Robot. Autom., pp. 4275–4280, 2002

    2002

  14. [14]

    Motion control of a tendon-based parallel manipulator using optimal tension distribution,

    S. Fang, D. Franitza, M. Torlo, F. Bekes, and M. Hiller, “Motion control of a tendon-based parallel manipulator using optimal tension distribution,”IEEE/ASME Trans. Mechatronics, vol. 9, no. 3, pp. 561– 568, 2004

    2004

  15. [15]

    Cable suspended planar robots with re- dundant cables: Controllers with positive tensions,

    S.-R. Oh and S. K. Agrawal, “Cable suspended planar robots with re- dundant cables: Controllers with positive tensions,”IEEE Trans. Robot., vol. 21, no. 3, pp. 457–465, 2005

    2005

  16. [16]

    Optimization of actuator forces in cable- based parallel manipulators using convex analysis,

    M. Hassan and A. Khajepour, “Optimization of actuator forces in cable- based parallel manipulators using convex analysis,”IEEE Trans. Robot., vol. 24, no. 3, pp. 736–740, 2008

    2008

  17. [17]

    Kinematic design of a 7-DOF cable-driven humanoid arm: a solution-in-nature approach,

    G. Yang, S. K. Mustafa, C. B. Pham, and S. H. Yeo, “Kinematic design of a 7-DOF cable-driven humanoid arm: a solution-in-nature approach,” inProc. IEEE/ASME Int. Conf. Adv. Int. Mechatronics, pp. 444–449, 2005

    2005

  18. [18]

    Reciprocal screw-based force-closure of an n-DOF open chain: Minimum number of cables required to fully constrain it,

    S. K. Mustafa and S. K. Agrawal, “Reciprocal screw-based force-closure of an n-DOF open chain: Minimum number of cables required to fully constrain it,” inProc. IEEE Int. Conf. Robot. Autom., pp. 3029–3034, 2011

    2011

  19. [19]

    Modeling and control of a cable driven modular snake robot,

    P. Racioppo and P. Ben-Tzvi, “Modeling and control of a cable driven modular snake robot,” inProc. IEEE Int. Conf. Contr. Technol. Appli- cations, pp. 468–473, 2017

    2017

  20. [20]

    Tension dis- tribution algorithm for planar mobile cable-driven parallel robots,

    T. Rasheed, P. Long, D. Marquez-Gamez, and S. Caro, “Tension dis- tribution algorithm for planar mobile cable-driven parallel robots,” in Cable-Driven Parallel Robots, ser. Mechanisms and Machine Science, C. Gosselin, P. Cardou, T. Bruckmann, and A. Pott, Eds., vol. 53, pp. 268–279. Springer International Publishing, 2018

    2018

  21. [21]

    A unified approach for motion and force control of robot manipulators: The operational space formulation,

    O. Khatib, “A unified approach for motion and force control of robot manipulators: The operational space formulation,”IEEE Trans. Robot. Autom., vol. 3, no. 1, pp. 43–53, 1987

    1987

  22. [22]

    Simulating the task- level control of human motion: a methodology and framework for implementation,

    V . D. Sapio, J. Warren, O. Khatib, and S. Delp, “Simulating the task- level control of human motion: a methodology and framework for implementation,”Visual Comput., vol. 21, no. 5, pp. 289–302, 2005

    2005

  23. [23]

    Per- turbed manipulability optimization in a distributed network of redundant robots,

    L. Jin, J. Zhang, X. Luo, M. Liu, S. Li, L. Xiao, and Z. Yang, “Per- turbed manipulability optimization in a distributed network of redundant robots,”IEEE Transactions on Industrial Electronics, vol. 68, no. 8, pp. 7209–7220, 2020

    2020

  24. [24]

    Ana- lytical inverse kinematic computation for 7-dof redundant manipulators with joint limits and its application to redundancy resolution,

    M. Shimizu, H. Kakuya, W.-K. Yoon, K. Kitagaki, and K. Kosuge, “Ana- lytical inverse kinematic computation for 7-dof redundant manipulators with joint limits and its application to redundancy resolution,”IEEE Transactions on robotics, vol. 24, no. 5, pp. 1131–1142, 2008

    2008

  25. [25]

    Modelling of cable wrapping phenomenon towards improved cable-driven mechanisms,

    M. C. Lei and D. Oetomo, “Modelling of cable wrapping phenomenon towards improved cable-driven mechanisms,” in2013 IEEE/ASME Inter- national Conference on Advanced Intelligent Mechatronics, pp. 649–655. IEEE, 2013

    2013

  26. [26]

    Bettering operation of robots by learning,

    S. Arimoto, S. Kawamura, and F. Miyazaki, “Bettering operation of robots by learning,”J. Robot. Syst., vol. 1, no. 2, pp. 123–140, 1984

    1984

  27. [27]

    Xu and Y

    J.-X. Xu and Y . Tan,Linear and Nonlinear Iterative Learning Control, ser. Lecture Notes in Control and Information Sciences, vol. 291. Springer-Verlag Berlin Heidelberg, 2003

    2003

  28. [28]

    A survey of iterative learning control,

    D. A. Bristow, M. Tharayil, and A. G. Alleyne, “A survey of iterative learning control,”IEEE Control Syst. Mag., vol. 26, no. 3, pp. 96–114, 2006

    2006

  29. [29]

    A unified data-driven design framework of optimality-based generalized iterative learning control,

    R. Chi, Z. Hou, B. Huang, and S. Jin, “A unified data-driven design framework of optimality-based generalized iterative learning control,” Computers & Chemical Engineering, vol. 77, pp. 10–23, 2015

    2015

  30. [30]

    CASPR: A comprehensive cable-robot analysis and simulation platform for the research of cable- driven parallel robots,

    D. Lau, J. Eden, Y . Tan, and D. Oetomo, “CASPR: A comprehensive cable-robot analysis and simulation platform for the research of cable- driven parallel robots,” inProc. IEEE/RSJ Int. Conf. Intell. Robot. Syst., pp. 3004–3011, 2016

    2016

  31. [31]

    Dual-space adaptive control of redundantly actuated cable-driven parallel robots,

    J. Lamaury, M. Gouttefarde, A. Chemori, and P. ´Elie Herv´e, “Dual-space adaptive control of redundantly actuated cable-driven parallel robots,” in Proc. IEEE/RSJ Int. Conf. Intell. Robot. Syst., pp. 4879–4886, 2013

    2013

  32. [32]

    Redundancy resolution and disturbance rejection via torque optimization in hybrid cable-driven robots

    R. Qi, A. Khajepour, and W. W. Melek, “Redundancy resolution and disturbance rejection via torque optimization in hybrid cable-driven robots.”

  33. [33]

    Tricriteria optimization-coordination motion of dual-redundant-robot manipulators for complex path planning,

    Z. Zhang, Y . Lin, S. Li, Y . Li, Z. Yu, and Y . Luo, “Tricriteria optimization-coordination motion of dual-redundant-robot manipulators for complex path planning,”IEEE Transactions on Control Systems Technology, vol. 26, no. 4, pp. 1345–1357, 2017

    2017

  34. [34]

    A varying-parameter convergent-differential neural network for solving joint-angular-drift problems of redundant robot manipulators,

    Z. Zhang, T. Fu, Z. Yan, L. Jin, L. Xiao, Y . Sun, Z. Yu, and Y . Li, “A varying-parameter convergent-differential neural network for solving joint-angular-drift problems of redundant robot manipulators,” IEEE/ASME Transactions on Mechatronics, vol. 23, no. 2, pp. 679–689, 2018

    2018

  35. [35]

    A varying parameter recurrent neural network for solving nonrepetitive motion problems of redundant robot manipulators,

    Z. Zhang and Z. Yan, “A varying parameter recurrent neural network for solving nonrepetitive motion problems of redundant robot manipulators,” IEEE Transactions on Control Systems Technology, vol. 27, no. 6, pp. 2680–2687, 2018

    2018

  36. [36]

    Compatible convex–nonconvex con- strained qp-based dual neural networks for motion planning of redundant robot manipulators,

    Z. Zhang, S. Chen, and S. Li, “Compatible convex–nonconvex con- strained qp-based dual neural networks for motion planning of redundant robot manipulators,”IEEE Transactions on Control Systems Technology, vol. 27, no. 3, pp. 1250–1258, 2018

    2018

  37. [37]

    Mutual- collision-avoidance scheme synthesized by neural networks for dual redundant robot manipulators executing cooperative tasks,

    Z. Zhang, L. Zheng, Z. Chen, L. Kong, and H. R. Karimi, “Mutual- collision-avoidance scheme synthesized by neural networks for dual redundant robot manipulators executing cooperative tasks,”IEEE trans- actions on neural networks and learning systems, vol. 32, no. 3, pp. 1052–1066, 2020

    2020

  38. [38]

    Inverse dynamics of multi- link cable-driven manipulators with the consideration of joint interaction forces and moments,

    D. Lau, D. Oetomo, and S. K. Halgamuge, “Inverse dynamics of multi- link cable-driven manipulators with the consideration of joint interaction forces and moments,”IEEE Trans. Robot., vol. 31, no. 2, pp. 479–488, 2015

    2015

  39. [39]

    Decoupled modeling and model predictive control of a hybrid cable-driven robot (hcdr),

    R. Qi, M. Rushton, A. Khajepour, and W. W. Melek, “Decoupled modeling and model predictive control of a hybrid cable-driven robot (hcdr),”Robotics and Autonomous Systems, vol. 118, pp. 1–12, 2019

    2019

  40. [40]

    Path following demonstration of a hybrid cable-driven parallel robot,

    M. Michelin, P. E. Herv ´e, O. Tempier, J. B. Izard, and M. Gouttefarde, “Path following demonstration of a hybrid cable-driven parallel robot,” inInternational Conference on Cable-Driven Parallel Robots, pp. 323–

  41. [41]

    A three-dimensional regression model of the shoulder rhythm,

    J. H. de Groot and R. Brand, “A three-dimensional regression model of the shoulder rhythm,”Clin. Biomech., vol. 16, no. 9, pp. 735–743, 2001

    2001

  42. [42]

    D. Lau, D. Oetomo, and S. K. Halgamuge, “Generalized modeling of multilink cable-driven manipulators with arbitrary routing using the 16 IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY , VOL. 31, NO. 6, NOVEMBER 2023 cable-routing matrix,”IEEE Trans. Robot., vol. 29, no. 5, pp. 1102–1113, 2013

    2023

  43. [43]

    Resolved motion rate control of manipulators and human prostheses,

    D. E. Whitney, “Resolved motion rate control of manipulators and human prostheses,”IEEE Trans. Man-Mach. Syst., vol. 10, no. 2, pp. 47–53, 1969

    1969

  44. [44]

    MY- OROBOTICS: a modular toolkit for legged locomotion research using musculoskeletal designs,

    H. G. Marques, C. Maufroy, A. Lenz, K. Dalamagkidis, U. Culha, M. Siee, P. Bremner, and the MYOROBOTICS Project Team, “MY- OROBOTICS: a modular toolkit for legged locomotion research using musculoskeletal designs,” inProc. 6th International Symposium on Adaptive Motion of Animals and Machines (AMAM’13), 2013

    2013

  45. [45]

    Dexterity measures with unilateral actuation constraints: the n+1 case,

    R. Kurtz and V . Hayward, “Dexterity measures with unilateral actuation constraints: the n+1 case,”Adv. Robot., vol. 9, no. 5, pp. 561–577, 1994

    1994

  46. [46]

    On fast computation of distance between line seg- ments,

    V . J. Lumelsky, “On fast computation of distance between line seg- ments,”Inform. Process. Lett., vol. 21, pp. 55–61, 1985

    1985

  47. [47]

    Adaptive iterative learning control for robot manipulators,

    A. Tayebi, “Adaptive iterative learning control for robot manipulators,” Automatica, vol. 40, no. 7, pp. 1195–1203, 2004

    2004

  48. [48]

    A data-driven constrained norm-optimal iterative learning control framework for LTI systems,

    P. Janssens, G. Pipeleers, and J. Swevers, “A data-driven constrained norm-optimal iterative learning control framework for LTI systems,” IEEE Trans. Control Syst. Technol., vol. 21, no. 2, pp. 546–551, 2013

    2013

  49. [49]

    An iterative learning control of robot manipulators,

    T.-Y . Kuc, K. Nam, and J. S. Lee, “An iterative learning control of robot manipulators,”IEEE Trans. Robot. Autom., vol. 7, no. 6, pp. 835–842, 1991

    1991

  50. [50]

    Least action principles and their application to constrained and task-level problems in robotics and biomechanics,

    V . D. Sapio, O. Khatib, and S. Delp, “Least action principles and their application to constrained and task-level problems in robotics and biomechanics,”Multibody Syst. Dyn., vol. 19, no. 3, pp. 303–322, 2008

    2008

  51. [51]

    On the limited memory BFGS method for large scale optimization,

    D. C. Liu and J. Nocedal, “On the limited memory BFGS method for large scale optimization,”Math. Program., vol. 45, no. 1, pp. 503–528, 1989

    1989

  52. [52]

    “direct search

    R. Hooke and T. A. Jeeves, ““direct search” solution of numerical and statistical problems,”J. ACM, vol. 8, no. 2, pp. 212–229, 1961

    1961

  53. [53]

    A new optimizer using particle swarm theory,

    R. Eberhart and J. Kennedy, “A new optimizer using particle swarm theory,” inProc. Sixth Int. Sypm. Micro Mach. Human Sci., pp. 39–43, 1995

    1995

  54. [54]

    Learning to control in operational space,

    J. Peters and S. Schaal, “Learning to control in operational space,”Int. J. Robot. Res., vol. 27, no. 2, pp. 197–212, 2008

    2008

  55. [55]

    Dynamic control of redundant manipulators,

    P. Hsu, J. Mauser, and S. Sastry, “Dynamic control of redundant manipulators,”Journal of Robotic Systems, vol. 6, no. 2, pp. 133–148, 1989

    1989

  56. [56]

    Perturbation theory and averaging,

    H. Khalil, “Perturbation theory and averaging,” inNonlinear Systems, 3rd ed., ch. 10, pp. 381–402. Pearson, 2001

    2001

  57. [57]

    Sufficient conditions for stabilization of sampled-data nonlinear systems via discrete-time approximations,

    D. Ne ˇsi´c, A. R. Teel, and P. V . Kokotovi ´c, “Sufficient conditions for stabilization of sampled-data nonlinear systems via discrete-time approximations,”Systems & Control Letters, vol. 38, no. 4-5, pp. 259– 270, 1999

    1999

  58. [58]

    qpOASES: A parametric active-set algorithm for quadratic program- ming,

    H. J. Ferreau, C. Kirches, A. Potschka, H. G. Bock, and M. Diehl, “qpOASES: A parametric active-set algorithm for quadratic program- ming,”Math. Program. Comput., vol. 6, no. 4, pp. 327–363, 2014

    2014

  59. [59]

    CASPR-ROS: A generalised cable robot software in ROS for hardware,

    J. Eden, C. Song, Y . Tan, D. Oetomo, and D. Lau, “CASPR-ROS: A generalised cable robot software in ROS for hardware,” inCable-Driven Parallel Robots, ser. Mechanisms and Machine Science, C. Gosselin, P. Cardou, T. Bruckmann, and A. Pott, Eds., vol. 53, pp. 50–61. Springer International Publishing, 2018. Dipankar BhattacharyaDipankar Bhattacharya (Member,...

    2018