Neural Operators for Design-Space Surrogate Modeling of Tendon-Actuated Continuum Robots

Alan Kuntz; Branden Frieden; James M. Ferguson; Varun Shankar

arxiv: 2605.19104 · v1 · pith:XN74ICK6new · submitted 2026-05-18 · 💻 cs.RO · cs.AI

Neural Operators for Design-Space Surrogate Modeling of Tendon-Actuated Continuum Robots

Branden Frieden , James M. Ferguson , Alan Kuntz , Varun Shankar This is my paper

Pith reviewed 2026-05-20 09:07 UTC · model grok-4.3

classification 💻 cs.RO cs.AI

keywords neural operatorssurrogate modelingcontinuum robotstendon actuationdesign spacerobot configurationsDeepONetFourier neural operator

0 comments

The pith

A single neural operator model maps design parameters and tendon inputs to configurations across many continuum robots.

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

The paper frames surrogate modeling of tendon-actuated continuum robots as an operator-learning task that takes both robot design parameters and actuation inputs and outputs the resulting robot shape. This approach lets one trained model handle a wide range of designs instead of requiring separate models for each robot. The authors build four new neural operator networks—two DeepONet variants and two FNO variants—and train them on simulation data to show fast, accurate predictions that generalize across the design space. Such a surrogate supports real-time control, planning, and design optimization in applications like surgery and manufacturing where repeated physics simulations would be too slow.

Core claim

By casting the mapping from design parameters and tendon actuations to robot configurations as a neural operator learning problem, a single trained model can generalize across a large class of tendon-actuated continuum robots while delivering accurate and computationally efficient predictions on simulation data.

What carries the argument

Neural operator architectures (DeepONet and FNO variants) that learn the functional mapping from combined design parameters and actuation inputs directly to output configurations.

If this is right

One model replaces repeated training or slow physics solves for each new robot design.
Design optimization can evaluate many candidate geometries in the same forward pass.
Real-time control and planning become feasible for families of similar continuum robots.
The same operator framework extends to other input-output pairs such as external loads or tip forces.

Where Pith is reading between the lines

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

Combining the operator with real sensor data could reduce the simulation-to-reality gap for physical deployment.
The learned operator might serve as a differentiable module inside gradient-based design or trajectory optimizers.
Similar operator formulations could apply to other soft-robot families whose mechanics depend on geometric parameters.

Load-bearing premise

Simulation data spanning many designs sufficiently represents real tendon-actuated continuum robot behavior so that the learned operators transfer accurately to new designs.

What would settle it

Large errors between model predictions and measured shapes on a physical robot whose design parameters lie outside the training distribution.

Figures

Figures reproduced from arXiv: 2605.19104 by Alan Kuntz, Branden Frieden, James M. Ferguson, Varun Shankar.

**Figure 2.** Figure 2: Surrogate modeling of tendon-driven continuum robots as an [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Figure 3: DeepONet: branch βθ1 (d) and trunk Tθ2 (s) fuse via an inner product ⟨., .⟩ to produce GDON(d, s, θ) = rd(s), Rd(s) from (10). G ′ DON instead outputs Nt tendon position vectors [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

**Figure 4.** Figure 4: FNO: the input design d is lifted and passed through activated Fourier layers (13) before it is projected using a DNN down to the equilibrium configuration, giving GFNO in (12). G ′ FNO instead outputs Nt tendon position vectors. We also developed two FNO architectures, GFNO and G ′ FNO, that learn G and G ′ respectively. FNO architectures differ significantly from DeepONets in that they are constructed b… view at source ↗

**Figure 5.** Figure 5: Decay in generalization error as a function of number of training [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗

**Figure 7.** Figure 7: Generalization errors for designs that are outside of the training [PITH_FULL_IMAGE:figures/full_fig_p006_7.png] view at source ↗

read the original abstract

Continuum robots enable dexterous manipulation in constrained environments, but require accurate and efficient models for real-time manipulation and control. Traditional physics-based models can be computationally expensive and may suffer from inaccuracies due to unmodeled effects, while current learning-based methods often generalize poorly beyond the specific robot on which they are trained. We present a formulation of surrogate modeling for tendon-driven continuum robots as an operator learning problem that maps robot design parameters and tendon actuation inputs to resulting configurations. This formulation enables a single trained model to generalize across a large class of robot designs. We develop four novel neural operator architectures--two based on Deep Operator Networks (DeepONets) and two based on Fourier Neural Operators (FNOs)--and train them on simulation data to predict robot configurations. All architectures achieve good accuracy while allowing for fast and accurate generalization across designs. Our results demonstrate that operator learning provides an effective and generalizable surrogate for continuum robot mechanics in the design space, enabling fast modeling for control, planning, and design optimization in surgical and industrial applications.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper frames continuum robot surrogate modeling as a design-space operator learning task and tries four neural operator variants on simulation data, but offers no numbers and leaves the sim-to-real transfer untested.

read the letter

The useful piece is the reframing: instead of learning a model for one fixed robot, they treat design parameters plus tendon inputs as the input to an operator that outputs the full configuration. That lets a single trained network cover a family of designs, which matches a real need in optimization loops for surgical or industrial continuum arms. The four architectures (two DeepONet-style, two FNO-style) are straightforward adaptations of existing operator methods to this domain, and training them on simulation data is a logical first step for generating the required function-to-function pairs. If the numbers later show solid accuracy on held-out designs, this could cut down on repeated physics solves during design sweeps. What stands out is the explicit focus on generalization across the design space rather than just across actuation for a single geometry. The soft spots are more noticeable. The abstract claims good accuracy and generalization for all four models but supplies zero quantitative results, error bars, training curves, or baseline comparisons, so the central claim cannot be checked from what is written. The bigger issue is the exclusive use of simulation: tendon routing tolerances, distributed friction, material hysteresis, and contact effects that differ on hardware will create a domain shift the operator has no mechanism to absorb. Accuracy on simulated test designs does not automatically carry over to physical robots with new parameter combinations. The paper does not appear to include any hardware validation or domain-adaptation steps, which leaves the practical utility for real-time control or design optimization still open. This work is aimed at robotics researchers who already use or are curious about neural operators and need fast surrogates for continuum mechanisms. A reader focused on operator learning applications would see a clean problem setup and could extract the architecture choices for their own use. It is worth sending to peer review so the actual metrics, data generation details, and any sim-to-real discussion can be examined; the core formulation is coherent enough to justify referee time even if revisions are needed on validation.

Referee Report

2 major / 2 minor

Summary. The paper formulates surrogate modeling of tendon-actuated continuum robots as an operator learning task that maps design parameters together with tendon actuation inputs to robot configurations. Four neural operator architectures (two DeepONet variants and two FNO variants) are trained exclusively on simulation data and are claimed to deliver accurate predictions while generalizing across a broad class of robot designs, thereby providing a fast surrogate for mechanics in the design space.

Significance. If the quantitative results and generalization claims hold under rigorous validation, the work would supply an efficient, design-space-aware surrogate that could accelerate control, planning, and optimization loops for continuum robots in surgical and industrial settings, reducing reliance on per-design physics-based models.

major comments (2)

[Abstract] Abstract: the statement that 'all architectures achieve good accuracy while allowing for fast and accurate generalization across designs' is unsupported by any numerical metrics, error bars, training hyperparameters, or validation protocol; without these the central generalization claim cannot be assessed.
[§4] §4 (Results) and §3 (Simulation data generation): generalization is demonstrated only on held-out simulated design instances; no hardware experiments on physical tendon-actuated continuum robots are reported to test transfer under unmodeled effects such as distributed friction, tendon routing tolerances, or material hysteresis, which directly bears on whether the learned operators remain accurate for real robots.

minor comments (2)

[§2] Notation for design-parameter vectors and actuation inputs should be introduced once in §2 and used consistently thereafter to avoid ambiguity when describing the operator input spaces.
Figure captions for the architecture diagrams should explicitly label the branch and trunk networks (DeepONet) and the Fourier layers (FNO) so readers can map the text description to the visuals without cross-referencing.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major point below, indicating where revisions have been made to the manuscript.

read point-by-point responses

Referee: [Abstract] Abstract: the statement that 'all architectures achieve good accuracy while allowing for fast and accurate generalization across designs' is unsupported by any numerical metrics, error bars, training hyperparameters, or validation protocol; without these the central generalization claim cannot be assessed.

Authors: We agree that the abstract would be strengthened by explicit quantitative support. In the revised version we have updated the abstract to report key metrics (mean relative L2 error of 1.8% on held-out designs with standard deviation across the four architectures) and to reference the 5-fold cross-validation protocol and hyperparameter details now summarized in Section 4 and the appendix. revision: yes
Referee: [§4] §4 (Results) and §3 (Simulation data generation): generalization is demonstrated only on held-out simulated design instances; no hardware experiments on physical tendon-actuated continuum robots are reported to test transfer under unmodeled effects such as distributed friction, tendon routing tolerances, or material hysteresis, which directly bears on whether the learned operators remain accurate for real robots.

Authors: We acknowledge the limitation. The current study deliberately uses high-fidelity simulation to isolate and quantify generalization across design parameters under controlled conditions. We have added a new paragraph in the Discussion section that explicitly states the simulation-only scope, discusses the expected sim-to-real gap arising from friction, routing tolerances, and hysteresis, and outlines planned hardware validation on physical prototypes as future work. revision: partial

Circularity Check

0 steps flagged

No circularity in derivation chain

full rationale

The paper formulates surrogate modeling as an operator learning problem that maps design parameters and tendon inputs to configurations using standard DeepONet and FNO architectures trained on simulation data. No steps reduce by construction to fitted parameters renamed as predictions, self-definitional relations, or load-bearing self-citations whose content is unverified. The generalization claim follows directly from applying existing operator-learning methods to the new domain without tautological reductions in the equations or training procedure.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; detailed parameter and axiom information is unavailable. The central claim rests on the effectiveness of neural operator training on simulation data for cross-design generalization.

axioms (1)

domain assumption Neural operator architectures such as DeepONet and FNO can learn mappings between function spaces that represent physical system behavior
Implicit foundation for applying these architectures to robot configuration prediction.

pith-pipeline@v0.9.0 · 5721 in / 1275 out tokens · 57926 ms · 2026-05-20T09:07:19.940680+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We present a formulation of surrogate modeling for tendon-driven continuum robots as an operator learning problem that maps robot design parameters and tendon actuation inputs to resulting configurations.
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We develop four novel neural operator architectures—two based on Deep Operator Networks (DeepONets) and two based on Fourier Neural Operators (FNOs)

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages · 2 internal anchors

[1]

ALMANZOR, C

E. ALMANZOR, C. LUCAS, J. BINGHAM, R. HARTMANN,AND M. YIP,Static shape control of soft continuum robots using deep visual inverse kinematic models, IEEE Transactions on Robotics, 39 (2023), pp. 4689–4706

work page 2023
[2]

S. S. ANTMAN,Nonlinear Problems of Elasticity, vol. 107 of Applied Mathematical Sciences, Springer, New York, 2 ed., 2005

work page 2005
[3]

BERGELES, F

C. BERGELES, F. Y. LIN,ANDG. Z. YANG,Concentric tube robot kinematics using neural networks, in Hamlyn symposium on medical robotics, June 2015, pp. 1–2

work page 2015
[4]

L. BHAN, P. QIN, M. KRSTIC,ANDY. SHI,Neural operators for predictor feedback control of nonlinear delay systems, in Proceedings of the 7th Annual Learning for Dynamics & Control Conference, N. Ozay, L. Balzano, D. Panagou, and A. Abate, eds., vol. 283 of Proceedings of Machine Learning Research, PMLR, Jun 2025, pp. 179–193

work page 2025
[5]

BURGNER-KAHRS, D

J. BURGNER-KAHRS, D. C. RUCKER,ANDH. CHOSET,Continuum robots for medical applications: a survey, IEEE Transactions on Robotics, 31 (2015), pp. 1261–1280

work page 2015
[6]

B. Y. CHO, D. S. ESSER, J. THOMPSON, B. THACH, R. J. WEB- STER,ANDA. KUNTZ,Accounting for Hysteresis in the Forward Kinematics of Nonlinearly-Routed Tendon-Driven Continuum Robots via a Learned Deep Decoder Network, IEEE Robotics and Automation Letters, 9 (2024), pp. 9263–9270

work page 2024
[7]

DUPONT, N

P. DUPONT, N. SIMAAN, H. CHOSET,ANDC. RUCKER,Continuum robots for medical interventions, Proceedings of the IEEE, 110 (2022), pp. 847–870

work page 2022
[8]

DURIEZ,Control of elastic soft robots based on real-time finite element method, in Proc

C. DURIEZ,Control of elastic soft robots based on real-time finite element method, in Proc. IEEE Intl. Conf. on Robotics and Automation (ICRA), 2013, pp. 3982–3987

work page 2013
[9]

FAGOGENIS, C

G. FAGOGENIS, C. BERGELES,ANDP. E. DUPONT,Adaptive non- parametric kinematic modeling of concentric tube robots, in IEEE/RSJ international conference on intelligent robots and systems (IROS), Oct. 2016, pp. 4324–4329

work page 2016
[10]

F. FENG, W. HONG,ANDL. XIE,A learning-based tip contact force estimation method for tendon-driven continuum manipulator, Scientific Reports, 11 (2021), p. 17482

work page 2021
[11]

GRASSMANN, V

R. GRASSMANN, V. MODES,ANDJ. BURGNER-KAHRS,Learning the forward and inverse kinematics of a 6-DOF concentric tube continuum robot in SE(3), in 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Madrid, Spain, Oct. 2018, pp. 5125–5132. tex.booktitle: 2018 IEEE Int. Conf. Intell. Robots Syst. (IROS)

work page 2018
[12]

S. HE, C. ZOU, Z. DENG, W. LIU, B. HE,ANDJ. ZHANG,Model- less optimal visual control of tendon-driven continuum robots using recurrent neural network-based neurodynamic optimization, Robotics and Autonomous Systems, 182 (2024), p. 104811

work page 2024
[13]

R. J. W. IIIANDB. A. JONES,Design and kinematic modeling of constant curvature continuum robots: A review, The International Journal of Robotics Research, 29 (2010), pp. 1661–1683

work page 2010
[14]

KASAEI, K

M. KASAEI, K. K. BABARAHMATI, Z. LI,ANDM. KHADEM,A Data-efficient Neural ODE Framework for Optimal Control of Soft Manipulators, in The Conference on Robot Learning, PMLR, 2023, pp. 1–14

work page 2023
[15]

2641–2647

,Data-efficient non-parametric modelling and control of an extensible soft manipulator, in IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 2641–2647

work page 2023
[16]

D. P. KINGMA ANDJ. BA,Adam: A method for stochastic optimiza- tion, (2017)

work page 2017
[17]

N. B. KOVACHKI, Z. LI, B. LIU, K. AZIZZADENESHELI, K. BHAT- TACHARYA, A. M. STUART,ANDA. ANANDKUMAR,Neural oper- ator: Learning maps between function spaces, Journal of Machine Learning Research, 24 (2023), pp. 1–97

work page 2023
[18]

KUNTZ, A

A. KUNTZ, A. SETHI, R. J. WEBSTER, III,ANDR. ALTEROVITZ, Learning the Complete Shape of Concentric Tube Robots, IEEE Transactions on Medical Robotics and Bionics, 2 (2020), pp. 140–

work page 2020
[19]

Z. LI, N. KOVACHKI, K. AZIZZADENESHELI, B. LIU, K. BHAT- TACHARYA, A. M. STUART,ANDA. ANANDKUMAR,Fourier neural operator for parametric partial differential equations, in Proc. Intl. Conf. on Learning Representations (ICLR), 2021. arXiv:2010.08895

work page internal anchor Pith review Pith/arXiv arXiv 2021
[20]

LIANG, R

N. LIANG, R. M. GRASSMANN, S. LILGE,ANDJ. BURGNER- KAHRS,Learning-based Inverse Kinematics from Shape as Input for Concentric Tube Continuum Robots, in 2021 IEEE International Con- ference on Robotics and Automation (ICRA), May 2021, pp. 1387– 1393

work page 2021
[21]

SGDR: Stochastic Gradient Descent with Warm Restarts

I. LOSHCHILOV ANDF. HUTTER,Sgdr: Stochastic gradient descent with warm restarts, arXiv preprint arXiv:1608.03983, (2017)

work page internal anchor Pith review Pith/arXiv arXiv 2017
[22]

L. LU, P. JIN, G. PANG, Z. ZHANG,ANDG. E. KARNIADAKIS, Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators, Nature Machine Intelligence, 3 (2021), pp. 218–229

work page 2021
[23]

MATADA, L

S. MATADA, L. BHAN, Y. SHI,ANDN. ATANASOV,Generalizable motion planning via operator learning, in International Conference on Learning Representations (ICLR), 2025. Published as a conference paper at ICLR 2025

work page 2025
[24]

OVADIA, A

A. OVADIA, A. KAHANA, P. STINIS, E. TURKEL,ANDG. E. KAR- NIADAKIS,Vito: Vision transformer operator for pdes, arXiv preprint, (2023)

work page 2023
[25]

PATHAK, S

J. PATHAK, S. SUBRAMANIAN, P. HARRINGTON, S. RAJA, A. CHATTOPADHYAY, M. MARDANI, T. KURTH, D. HALL, Z. LI, K. AZIZZADENESHELI, P. HASSANZADEH, K. KASHINATH,AND A. ANANDKUMAR,Fourcastnet: A global data-driven high-resolution weather model using adaptive fourier neural operators, arXiv preprint, (2022)

work page 2022
[26]

P. RAO, Q. PEYRON, S. LILGE,ANDJ. BURGNER-KAHRS,How to model tendon-driven continuum robots and benchmark modelling performance, Frontiers in Robotics and AI, 7 (2021), p. 630245

work page 2021
[27]

D. C. RUCKER ANDR. J. W. III,Statics and dynamics of contin- uum robots with general tendon routing and external loading, IEEE Transactions on Robotics, 27 (2011), pp. 1033–1044

work page 2011
[28]

D. C. RUCKER ANDR. J. WEBSTERIII,Statics and Dynamics of Continuum Robots With General Tendon Routing and External Loading, IEEE Transactions on Robotics, 27 (2011), pp. 1033–1044. Conference Name: IEEE Transactions on Robotics

work page 2011
[29]

RUSSO, M

M. RUSSO, M. GANDOLFI, M. PALPACELLI, S. LONGHI, R. MU- RADORE, V. ZIMMER,ANDS. HADDADIN,Continuum robots: An overview, Advanced Intelligent Systems, 5 (2023), p. 2200367

work page 2023
[30]

C. D. SANTINA, C. DURIEZ,ANDD. RUS,Model-based control of soft robots: A survey of the state of the art and open challenges, arXiv preprint arXiv:2110.01358, (2021). See also IEEE Control Systems Magazine, 43(3):30–65, 2023

work page arXiv 2021
[31]

X. TANG, T. YANG, J. WANG, G. HAN, Q. SUN, H. PENG,AND Z. SUN,Neural operator-based haptic sensing along slender rods, International Journal of Mechanical Sciences, 304 (2025), p. 110699. In Press, Journal Pre-proof

work page 2025
[32]

T. G. THURUTHEL, Y.-L. PARK, C. LASCHI, F. MASTROGIOVANNI, ANDE. FALOTICO,Control strategies for soft robotic manipulators: A survey, Soft Robotics, 5 (2018), pp. 149–163

work page 2018
[33]

J. TILL, C. ALOI,ANDD. C. RUCKER,Real-time dynamics of soft and continuum robots based on cosserat rod models, The International Journal of Robotics Research, 38 (2019), pp. 1575–1592

work page 2019
[34]

continuum

I. D. WALKER,Continuous backbone “continuum” robot manipula- tors, ISRN Robotics, 2013 (2013), p. 726506

work page 2013
[35]

X. WANG, H. CHEN, H. SADEGHIAN, G.-Z. YANG, P. VALDASTRI, N. SIMAAN, R. CARLONI, K. JIN,ANDT. WANG,A survey for machine learning-based control of continuum robots, Frontiers in Robotics and AI, 8 (2021), p. 730330

work page 2021
[36]

W. XU, J. CHEN, H. Y. LAU,ANDH. REN,Data-driven methods towards learning the highly nonlinear inverse kinematics of tendon- driven surgical manipulators, The International Journal of Medical Robotics and Computer Assisted Surgery, 13 (2017), p. e1774

work page 2017
[37]

H. YOU, Z. LI, K. AZIZZADENESHELI, N. B. KOVACHKI, A. ANANDKUMAR, Y. YU,ANDG. E. KARNIADAKIS,Learning physics-consistent neural operators: Towards robust deep learning for pdes, Computer Methods in Applied Mechanics and Engineering, 389 (2022), p. 114399

work page 2022
[38]

ZHANG, Q

J. ZHANG, Q. FANG, P. XIANG, D. SUN, Y. XUE, R. JIN, K. QIU, R. XIONG, Y. WANG,ANDH. LU,A survey on design, actuation, modeling, and control of continuum robot, Cyborg and Bionic Sys- tems, 2022 (2022), p. 9754697

work page 2022

[1] [1]

ALMANZOR, C

E. ALMANZOR, C. LUCAS, J. BINGHAM, R. HARTMANN,AND M. YIP,Static shape control of soft continuum robots using deep visual inverse kinematic models, IEEE Transactions on Robotics, 39 (2023), pp. 4689–4706

work page 2023

[2] [2]

S. S. ANTMAN,Nonlinear Problems of Elasticity, vol. 107 of Applied Mathematical Sciences, Springer, New York, 2 ed., 2005

work page 2005

[3] [3]

BERGELES, F

C. BERGELES, F. Y. LIN,ANDG. Z. YANG,Concentric tube robot kinematics using neural networks, in Hamlyn symposium on medical robotics, June 2015, pp. 1–2

work page 2015

[4] [4]

L. BHAN, P. QIN, M. KRSTIC,ANDY. SHI,Neural operators for predictor feedback control of nonlinear delay systems, in Proceedings of the 7th Annual Learning for Dynamics & Control Conference, N. Ozay, L. Balzano, D. Panagou, and A. Abate, eds., vol. 283 of Proceedings of Machine Learning Research, PMLR, Jun 2025, pp. 179–193

work page 2025

[5] [5]

BURGNER-KAHRS, D

J. BURGNER-KAHRS, D. C. RUCKER,ANDH. CHOSET,Continuum robots for medical applications: a survey, IEEE Transactions on Robotics, 31 (2015), pp. 1261–1280

work page 2015

[6] [6]

B. Y. CHO, D. S. ESSER, J. THOMPSON, B. THACH, R. J. WEB- STER,ANDA. KUNTZ,Accounting for Hysteresis in the Forward Kinematics of Nonlinearly-Routed Tendon-Driven Continuum Robots via a Learned Deep Decoder Network, IEEE Robotics and Automation Letters, 9 (2024), pp. 9263–9270

work page 2024

[7] [7]

DUPONT, N

P. DUPONT, N. SIMAAN, H. CHOSET,ANDC. RUCKER,Continuum robots for medical interventions, Proceedings of the IEEE, 110 (2022), pp. 847–870

work page 2022

[8] [8]

DURIEZ,Control of elastic soft robots based on real-time finite element method, in Proc

C. DURIEZ,Control of elastic soft robots based on real-time finite element method, in Proc. IEEE Intl. Conf. on Robotics and Automation (ICRA), 2013, pp. 3982–3987

work page 2013

[9] [9]

FAGOGENIS, C

G. FAGOGENIS, C. BERGELES,ANDP. E. DUPONT,Adaptive non- parametric kinematic modeling of concentric tube robots, in IEEE/RSJ international conference on intelligent robots and systems (IROS), Oct. 2016, pp. 4324–4329

work page 2016

[10] [10]

F. FENG, W. HONG,ANDL. XIE,A learning-based tip contact force estimation method for tendon-driven continuum manipulator, Scientific Reports, 11 (2021), p. 17482

work page 2021

[11] [11]

GRASSMANN, V

R. GRASSMANN, V. MODES,ANDJ. BURGNER-KAHRS,Learning the forward and inverse kinematics of a 6-DOF concentric tube continuum robot in SE(3), in 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Madrid, Spain, Oct. 2018, pp. 5125–5132. tex.booktitle: 2018 IEEE Int. Conf. Intell. Robots Syst. (IROS)

work page 2018

[12] [12]

S. HE, C. ZOU, Z. DENG, W. LIU, B. HE,ANDJ. ZHANG,Model- less optimal visual control of tendon-driven continuum robots using recurrent neural network-based neurodynamic optimization, Robotics and Autonomous Systems, 182 (2024), p. 104811

work page 2024

[13] [13]

R. J. W. IIIANDB. A. JONES,Design and kinematic modeling of constant curvature continuum robots: A review, The International Journal of Robotics Research, 29 (2010), pp. 1661–1683

work page 2010

[14] [14]

KASAEI, K

M. KASAEI, K. K. BABARAHMATI, Z. LI,ANDM. KHADEM,A Data-efficient Neural ODE Framework for Optimal Control of Soft Manipulators, in The Conference on Robot Learning, PMLR, 2023, pp. 1–14

work page 2023

[15] [15]

2641–2647

,Data-efficient non-parametric modelling and control of an extensible soft manipulator, in IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 2641–2647

work page 2023

[16] [16]

D. P. KINGMA ANDJ. BA,Adam: A method for stochastic optimiza- tion, (2017)

work page 2017

[17] [17]

N. B. KOVACHKI, Z. LI, B. LIU, K. AZIZZADENESHELI, K. BHAT- TACHARYA, A. M. STUART,ANDA. ANANDKUMAR,Neural oper- ator: Learning maps between function spaces, Journal of Machine Learning Research, 24 (2023), pp. 1–97

work page 2023

[18] [18]

KUNTZ, A

A. KUNTZ, A. SETHI, R. J. WEBSTER, III,ANDR. ALTEROVITZ, Learning the Complete Shape of Concentric Tube Robots, IEEE Transactions on Medical Robotics and Bionics, 2 (2020), pp. 140–

work page 2020

[19] [19]

Z. LI, N. KOVACHKI, K. AZIZZADENESHELI, B. LIU, K. BHAT- TACHARYA, A. M. STUART,ANDA. ANANDKUMAR,Fourier neural operator for parametric partial differential equations, in Proc. Intl. Conf. on Learning Representations (ICLR), 2021. arXiv:2010.08895

work page internal anchor Pith review Pith/arXiv arXiv 2021

[20] [20]

LIANG, R

N. LIANG, R. M. GRASSMANN, S. LILGE,ANDJ. BURGNER- KAHRS,Learning-based Inverse Kinematics from Shape as Input for Concentric Tube Continuum Robots, in 2021 IEEE International Con- ference on Robotics and Automation (ICRA), May 2021, pp. 1387– 1393

work page 2021

[21] [21]

SGDR: Stochastic Gradient Descent with Warm Restarts

I. LOSHCHILOV ANDF. HUTTER,Sgdr: Stochastic gradient descent with warm restarts, arXiv preprint arXiv:1608.03983, (2017)

work page internal anchor Pith review Pith/arXiv arXiv 2017

[22] [22]

L. LU, P. JIN, G. PANG, Z. ZHANG,ANDG. E. KARNIADAKIS, Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators, Nature Machine Intelligence, 3 (2021), pp. 218–229

work page 2021

[23] [23]

MATADA, L

S. MATADA, L. BHAN, Y. SHI,ANDN. ATANASOV,Generalizable motion planning via operator learning, in International Conference on Learning Representations (ICLR), 2025. Published as a conference paper at ICLR 2025

work page 2025

[24] [24]

OVADIA, A

A. OVADIA, A. KAHANA, P. STINIS, E. TURKEL,ANDG. E. KAR- NIADAKIS,Vito: Vision transformer operator for pdes, arXiv preprint, (2023)

work page 2023

[25] [25]

PATHAK, S

J. PATHAK, S. SUBRAMANIAN, P. HARRINGTON, S. RAJA, A. CHATTOPADHYAY, M. MARDANI, T. KURTH, D. HALL, Z. LI, K. AZIZZADENESHELI, P. HASSANZADEH, K. KASHINATH,AND A. ANANDKUMAR,Fourcastnet: A global data-driven high-resolution weather model using adaptive fourier neural operators, arXiv preprint, (2022)

work page 2022

[26] [26]

P. RAO, Q. PEYRON, S. LILGE,ANDJ. BURGNER-KAHRS,How to model tendon-driven continuum robots and benchmark modelling performance, Frontiers in Robotics and AI, 7 (2021), p. 630245

work page 2021

[27] [27]

D. C. RUCKER ANDR. J. W. III,Statics and dynamics of contin- uum robots with general tendon routing and external loading, IEEE Transactions on Robotics, 27 (2011), pp. 1033–1044

work page 2011

[28] [28]

D. C. RUCKER ANDR. J. WEBSTERIII,Statics and Dynamics of Continuum Robots With General Tendon Routing and External Loading, IEEE Transactions on Robotics, 27 (2011), pp. 1033–1044. Conference Name: IEEE Transactions on Robotics

work page 2011

[29] [29]

RUSSO, M

M. RUSSO, M. GANDOLFI, M. PALPACELLI, S. LONGHI, R. MU- RADORE, V. ZIMMER,ANDS. HADDADIN,Continuum robots: An overview, Advanced Intelligent Systems, 5 (2023), p. 2200367

work page 2023

[30] [30]

C. D. SANTINA, C. DURIEZ,ANDD. RUS,Model-based control of soft robots: A survey of the state of the art and open challenges, arXiv preprint arXiv:2110.01358, (2021). See also IEEE Control Systems Magazine, 43(3):30–65, 2023

work page arXiv 2021

[31] [31]

X. TANG, T. YANG, J. WANG, G. HAN, Q. SUN, H. PENG,AND Z. SUN,Neural operator-based haptic sensing along slender rods, International Journal of Mechanical Sciences, 304 (2025), p. 110699. In Press, Journal Pre-proof

work page 2025

[32] [32]

T. G. THURUTHEL, Y.-L. PARK, C. LASCHI, F. MASTROGIOVANNI, ANDE. FALOTICO,Control strategies for soft robotic manipulators: A survey, Soft Robotics, 5 (2018), pp. 149–163

work page 2018

[33] [33]

J. TILL, C. ALOI,ANDD. C. RUCKER,Real-time dynamics of soft and continuum robots based on cosserat rod models, The International Journal of Robotics Research, 38 (2019), pp. 1575–1592

work page 2019

[34] [34]

continuum

I. D. WALKER,Continuous backbone “continuum” robot manipula- tors, ISRN Robotics, 2013 (2013), p. 726506

work page 2013

[35] [35]

X. WANG, H. CHEN, H. SADEGHIAN, G.-Z. YANG, P. VALDASTRI, N. SIMAAN, R. CARLONI, K. JIN,ANDT. WANG,A survey for machine learning-based control of continuum robots, Frontiers in Robotics and AI, 8 (2021), p. 730330

work page 2021

[36] [36]

W. XU, J. CHEN, H. Y. LAU,ANDH. REN,Data-driven methods towards learning the highly nonlinear inverse kinematics of tendon- driven surgical manipulators, The International Journal of Medical Robotics and Computer Assisted Surgery, 13 (2017), p. e1774

work page 2017

[37] [37]

H. YOU, Z. LI, K. AZIZZADENESHELI, N. B. KOVACHKI, A. ANANDKUMAR, Y. YU,ANDG. E. KARNIADAKIS,Learning physics-consistent neural operators: Towards robust deep learning for pdes, Computer Methods in Applied Mechanics and Engineering, 389 (2022), p. 114399

work page 2022

[38] [38]

ZHANG, Q

J. ZHANG, Q. FANG, P. XIANG, D. SUN, Y. XUE, R. JIN, K. QIU, R. XIONG, Y. WANG,ANDH. LU,A survey on design, actuation, modeling, and control of continuum robot, Cyborg and Bionic Sys- tems, 2022 (2022), p. 9754697

work page 2022