Pith. sign in

REVIEW 3 major objections 5 minor 53 references

ATP: Anatomical Torque with Passivity-based Control Framework for Safe Upper-Limb Exoskeleton Assistance

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A unified framework turns anatomical muscle torques into safe, trajectory-free assistance for upper-limb exoskeletons.

desk verdict Solid integration of learned muscle-derived torque references with passivity-based delivery; the tracking guarantee is shakier than the abstract suggests. read the letter →

arxiv 2608.05723 v1 pith:YF5ZF3M6 submitted 2026-08-06 cs.RO

Yu Chen , Gong Chen , Xiang Li This is my paper · ORCID
classification cs.RO
keywords upper-limbexoskeletonanatomicalassistancepassivity-basedcontrolenergytankmusculoskeletalmodelreinforcementlearningtorquerefinementelectromyographyevaluation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes ATP, a framework that aims to let an upper-limb exoskeleton assist natural, unscripted arm movements by converting anatomical, muscle-derived torques into physical support. It claims to be the first exoskeleton assistance framework to generate anatomical assistance for multi-joint, complex, nonperiodic upper-limb movements without relying on predefined motion trajectories. The pipeline trains a unified muscle controller in a scalable musculoskeletal simulation, refines its torque output online using safety and comfort signals, and delivers the result through a backstepping controller with energy tanks that preserve passivity. If the central claim holds, exoskeletons could provide movement-dependent, task-general support in daily life rather than only for scripted tasks. Simulation, real-hardware, and EMG experiments with five participants report accurate torque tracking and up to 48% reduction in target-muscle activity.

What carries the argument

The load-bearing mechanism is the three-stage ATP chain. First, a unified muscle controller, trained by reinforcement learning in a GPU-accelerated musculoskeletal simulation, maps wearer kinematics to anatomical torque references and generalizes across diverse movements. Second, an online torque-refinement optimization repeatedly minimizes the distance from the reference while penalizing torque-rate and a learned anomaly score, which suppresses tendon-induced spikes and reshapes assistance during collisions or near kinematic singularities. Third, an interaction-torque controller built with backstepping for the cable-driven series-elastic exoskeleton uses two virtual energy tanks—reservoirs that store or release interaction energy—so that when a tank is depleted the controller relaxes the tracking objective to keep the human-robot port passive, and resumes tracking once replenished. Theorem 1 gives asymptotic stability and accurate torque tracking under the condition that the environment dissipates energy (inequality (53)) and the gain and parameter conditions (37)–(39) hold.

What would settle it

Have a participant deliberately push against the assisted joints so that interaction power $\dot{q}^T\tau_e$ is positive over a sustained interval, and record whether the closed loop remains passive and torque tracking is preserved. If the system still tracks torque while net energy flows from the human into the robot, the dissipative-environment inequality (53) used in Theorem 1 is violated; if tracking degrades exactly when tank energy is exhausted, the guarantee holds only under the paper's stated assumption.

Watch

Extended reading notes

Core claim

The central discovery, stated on the paper's own terms, is that a unified reinforcement-learning muscle controller trained in a scalable musculoskeletal simulation can produce real-time anatomical joint-torque references that generalize across many upper-limb motions, and that these references can be made safe and deliverable by online refinement plus a passivity-preserving interaction-torque controller. The controller tracks the refined torque on a cable-driven series-elastic exoskeleton without constraining the wearer to a preset trajectory, and the energy-tank corrections guarantee passivity and resume torque tracking after the tank is replenished. The pilot EMG evidence supports the claim that assistance reduces the activity of primary target muscles by up to 48% relative to moving without the exoskeleton, and reduces it relative to gravity compensation and open-loop assistance in the tested tasks.

Load-bearing premise

The safety and stability guarantee assumes the wearer's arm and surroundings only absorb energy and never actively push energy back into the exoskeleton, so a person who deliberately resists or drives the motion lies outside the guarantee.

Editorial extensions

If this is right

  • Upper-limb exoskeletons can deliver assist-as-needed torque support during unscripted, multi-joint movements without restricting the wearer's motion freedom.
  • A muscle controller trained jointly on diverse motion datasets retains most of its per-task tracking performance and generalizes to real-time, unseen movements measured by IMUs.
  • The online refinement module can detect and respond to anomalous interaction conditions, reducing assistance during simulated collisions and generating corrective torque near kinematic singularities.
  • The energy-tank controller preserves passivity even when tracking is temporarily suspended, and automatically returns to accurate torque tracking after the tank is replenished.
  • In the pilot EMG study, ATP reduced primary target-muscle activity by up to 48% in a dynamic multi-joint task compared with moving without the exoskeleton, and matched or improved on gravity-compensation and open-loop assistance.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the passivity proof relies on a dissipative human/environment, field deployments would need to detect active user effort or otherwise guarantee safe behavior when the wearer pushes energy into the system.
  • The same learned-muscle-reference plus energy-tank delivery pattern could transfer to other compliant wearable robots; a direct next test is whether metabolic cost, not just EMG, decreases as assistance is scaled.
  • Because the anomaly score guides refinement in only two tested scenarios, the framework invites extension to other anomalies, such as payload handling or involuntary spasms, which the paper explicitly leaves for future work.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes ATP, a three-stage framework for upper-limb exoskeleton assistance: a reinforcement-learned musculoskeletal muscle controller that generates anatomical reference torques from a MuJoCo/MyoArm model, an online torque-refinement module that smooths the reference and incorporates a learned anomaly score, and an interaction-torque controller with energy-tank-based passivity preservation for a cable-driven SEA exoskeleton. The authors report simulation results, real-hardware torque-tracking experiments on two direct-drive shoulder joints and one cable-driven elbow joint, and an EMG study with five participants showing reduced target-muscle activity during static and dynamic tasks, with reductions up to 48% relative to unassisted movement. The central claims are that ATP generalizes to complex nonperiodic upper-limb movements without predefined trajectories and that the interaction-torque controller provides rigorous closed-loop torque-tracking guarantees subject to passivity.

Significance. If the results hold, the paper would make a useful contribution by combining learned biomechanical torque references with passivity-based interaction control for upper-limb exoskeletons. The open-sourced GPU-accelerated musculoskeletal training environment, the real-world torque-tracking RMSEs of 0.14-0.27 Nm, the explicit passivity demonstrations during tank switching, and the ablation of anomaly-guided torque refinement are concrete strengths that go beyond a purely simulation-based study. However, the paper's theoretical torque-tracking guarantee currently depends on an unvalidated environment-passivity assumption, and the EMG evaluation lacks inferential statistics. These issues do not invalidate the empirical results, but they require the theoretical claims to be restated or supported with additional evidence.

major comments (3)
  1. [Section V, Theorem 1 and Eq. (53)] The proof of Theorem 1 relies on Eq. (53), which asserts that the human/environment is dissipative: Hdot_env(x_env) <= -qdot^T tau_e, justified only by the statement 'Dissipation can always be presumed in the environment.' This is a behavioral assumption about the wearer, not a property of the robot controller, and it is not measured or enforced in the experiments. Indeed, the tank-replenishment phases in Figs. 14 and 15 (green shaded regions) show the wearer moving the joint so that the interaction torque is aligned with the joint rotation, i.e., qdot^T tau_e > 0, which means the wearer is injecting energy into the exoskeleton. Under Eq. (53), such an interaction would require the human arm to be dissipating mechanical energy at the same moment. If the wearer is instead an active energy source, Vdot in Eq. (52) is not necessarily negative, so Theorem 1 does not establish asymptotic stability or torque-tracking convergence. The observed 'resumes tracking after tank replenishment' is therefore an empirical result, not a theoretical guarantee. The authors should either measure or enforce environment passivity, or explicitly restrict the claim to passive environments and soften the abstract/introduction statement that 'rigorous theoretical analysis guarantees closed-loop torque tracking subject to passivity.'
  2. [Section V, Proposition 1] The proof of Proposition 1 asserts, immediately before Eq. (23), that 'By choosing a sufficiently large beta_q and a correspondingly larger K_q, there exists a finite mu such that lim sup_{t->inf} ||M(q) qddot~|| <= mu whenever ||tau_tilde_f|| <= zeta.' This bound on ||M(q) qddot~|| is load-bearing for the subsequent uniform ultimate boundedness argument in Eqs. (24)-(27), but no derivation is provided. The bound should follow from the observer error dynamics (22) with explicit conditions on beta_q, K_q, and the boundedness properties of M(q), C(qdot,q), and g(q); as written, the proof leaves a gap between the stability of qtilde and the required bound on the second derivative. Please supply this derivation or state the additional assumptions needed.
  3. [Section VII-C, Fig. 19] The EMG study is presented as a central validation of the assistance benefit, but no inferential statistics are reported. The text says the data were 'statistically analyzed across all trials,' yet the results in Fig. 19 are summarized only as mean reductions (e.g., 64%, 45%, 59%, 33%, 48%, 47%) for a cohort of five participants. Because the static-task results are very similar between ATP and gravity compensation, and the dynamic-task effects vary across muscles and tasks, the claim that ATP reduces target-muscle activity compared with the baselines needs repeated-measures statistical analysis (or at least confidence intervals and effect sizes) to be supported. Without such analysis, the reported percentages may not be robust to individual-subject variability.
minor comments (5)
  1. [Section IV, Eq. (16)] The anomaly-score dynamics s(t+1) = s(t) + (partial f_a / partial tau_e)^T u_d Delta t assume that the diffusion-model-based anomaly score is differentiable with respect to tau_e; please clarify whether this gradient is computed analytically or numerically, and whether s is treated as a scalar throughout.
  2. [Section V, Eqs. (41) and (44)] The tank dynamics contain divisions by t_f and t_o. If the tanks can reach exactly zero energy, the right-hand sides are singular. The definitions of L_{f,o} and delta_{f,o} suggest the tanks are prevented from reaching zero, but this should be stated explicitly, together with the initialization of t_f and t_o.
  3. [Table V] The entries for Lafan1 and Ours have no standard deviations, while the other datasets do; the meaning of the 'Ratio' column (e.g., 1.55 (64.5%)) should be explained more clearly, since the percentage appears to be a performance-retention measure rather than a ratio.
  4. [Section VI] The ethics approval text contains the placeholder 'XXX' ('approved by the ethics committee of XXX'); the institution should be named.
  5. [Abstract and Section VIII] The source code is mentioned as 'available at ATP_muscle_controller,' but no URL or repository identifier is given; please provide a complete reference.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the torque-tracking and passivity results rest on explicit dissipativity assumptions and algebraic tank design, while the EMG outcomes are measured rather than model-derived.

full rationale

The derivation chain is not circular. The muscle controller is trained in a GPU-accelerated MuJoCo/MyoArm environment to minimize a tracking reward (Eqs. 6-8), and the resulting torque reference is a model output; the reported EMG reductions in Section VII-C are measured from five participants, not computed from the model or from the assistance scaling. The passivity/torque-tracking argument in Theorem 1 is a standard Lyapunov and energy-tank construction: Eq. (53), Hdot_env(x_env) <= -qdot^T tau_e, is an explicit dissipativity assumption on the human/environment, and the theorem is stated conditionally as 'if the passivity is not violated.' The tank corrections (43) and (48) together with Table III verify Hdot <= qdot^T tau_e algebraically, so the passivity guarantee is a design verification rather than a renamed input. The 'resumes tracking after energy-tank replenishment' behavior is demonstrated experimentally in Figs. 14-15, and is therefore an empirical result rather than a prediction generated from Eq. (53). The anomaly detector and intention predictor from the authors' prior work [46] are used as off-the-shelf components and are not relied upon to prove the central torque-tracking guarantee or the EMG outcomes; their reuse is a normal component citation, not a load-bearing self-citation. No fitted parameter is renamed as a prediction, and no quantity is defined in terms of the quantity it is supposed to predict. The main caveat, that Eq. (53) is a behavioral assumption about the wearer that the tank-replenishment experiments do not directly measure, is a soundness and applicability concern rather than a circularity in the derivation chain.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The framework rests on standard manipulator equations, a passivity assumption about the human environment that is not measured, and an unvalidated equivalence between the MyoArm simulation and the wearer. Several crucial tuning values (MPC weights, tank thresholds, real-world controller gains, friction bound zeta) are not reported.

free parameters (5)
  • EMG assistance scaling factor = 30% of tau_d baseline for 70 kg, scaled by body weight
    Set in Section VII.C without optimization or physiological justification; the magnitude of assistance directly affects the reported EMG reductions.
  • Torque-refinement look-ahead interval = 10 ms
    Empirically selected in Section VII.C to compensate inference latency; a free tuning choice affecting responsiveness and comfort.
  • Muscle controller reward weights = sigma_p=0.3, sigma_pdot=30, lambda_a=1.5, alpha_p=0.001, alpha_pdot=0.001
    Hand-chosen training hyperparameters in Table IV; they determine the reference torque quality and spiking behavior.
  • Passivity controller gains = K_p=0.5, K_d=0.05, K_v=5, K_s=2.0, K_theta=1.0, alpha_theta=5.0, K_e=4.0, beta_e=30, K=2.0 Nm/rad (simulation)
    Simulation gains in Section VII.B satisfy (37)-(39); real-world gain values are not disclosed, so the reported experimental behavior cannot be exactly reproduced.
  • Torque-refinement MPC weights/horizon = Not reported
    Q, R, N_p, T, U in (16) are defined but never assigned values; the refinement behavior is therefore only qualitatively described.
assumptions (6)
  • standard math The robot dynamics satisfy M(q) positive definite/symmetric, Mdot-2C skew-symmetric, and B, K constant diagonal positive definite.
    Stated as conditions in Section III and used throughout the backstepping and passivity proofs.
  • domain assumption The human/environment is passive and dissipative: Hdot_env(x_env) <= -qdot^T tau_e.
    Equation (53) in Section V is indispensable for Theorem 1's asymptotic stability; it is presumed, not verified in experiments, and may fail for an actively moving wearer.
  • domain assumption The MyoArm musculoskeletal model is a valid stand-in for the wearer's upper limb, and the learned muscle torques constitute an anatomical reference.
    The whole 'anatomical' claim depends on this equivalence; no comparison to measured human joint torques or muscle activity is provided.
  • domain assumption The inverse kinematic mapping q_h = f^{-1}(q_m) exists and the Jacobian can be inverted except at the noted singularity.
    Used in Section IV to map between exoskeleton and musculoskeletal coordinates; singular configurations are handled in refinement, but the mapping is not otherwise validated.
  • domain assumption The friction estimation error is bounded: ||tau_tilde_f|| <= zeta.
    Required for Proposition 1's UUB radius and for the damping condition k_g > zeta; the bound zeta is never quantified.
  • domain assumption The anomaly score from [46] is a valid safety/comfort signal for online refinement.
    The torque refinement cost includes s^2, but the anomaly detector is trained on transparent-mode data from healthy subjects and is treated as ground truth for comfort.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ATP: Anatomical Torque with Passivity-based Control Framework for Safe Upper-Limb Exoskeleton Assistance." pith.science (2026). https://pith.science/paper/YF5ZF3M6

@misc{pith2026260805723,
  author       = {Pith},
  title        = {Pith review of: ATP: Anatomical Torque with Passivity-based Control Framework for Safe Upper-Limb Exoskeleton Assistance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YF5ZF3M6}},
  note         = {Machine review of arXiv:2608.05723}
}
read the original abstract

Providing assistance across diverse movements is a central objective of exoskeletons, and anatomical knowledge can enable responsive support that generalizes across tasks. However, anatomical assistance has mainly been studied for lower-limb exoskeletons, where periodic, weight-bearing motions impose lower demands on torque precision. Extending such assistance to complex, nonperiodic upper-limb movements remains challenging. This paper proposes Anatomical Torque with Passivity-Based Control (ATP) for safe upper-limb exoskeleton assistance. First, a scalable musculoskeletal simulation framework trains a unified reinforcement-learning muscle controller that generalizes across upper-limb movements and generates anatomical reference torques without complex biomechanical computations. Second, an online torque-refinement scheme adapts the reference to diverse movements, suppresses tendon-induced spikes, and incorporates a learned anomaly score for safe and comfortable assistance. Third, an interaction torque controller delivers assistance through a cable-driven compliant exoskeleton without constraining motion to predefined trajectories, while an energy tank preserves passivity with theoretical guarantees on torque tracking and system passivity. Simulations and real-world experiments show accurate tracking of long-duration motion sequences and generalization to real-time human movements. The controller achieves accurate torque tracking while preserving passivity and resumes tracking after energy-tank replenishment. An EMG study with five participants further shows reduced target-muscle activity during static and dynamic tasks compared with gravity compensation and open-loop assistance, with reductions of up to 48% relative to movement without the exoskeleton in a dynamic multi-joint task.

Figures

Figures reproduced from arXiv: 2608.05723 by the authors.

Figure 1
Figure 1. Illustration of the anatomical assistance with exoskeleton. The proposed method enables the upper-limb exoskeleton to provide anatomical assistance across diverse upper-limb movements. technology continues to evolve, bio-inspired approaches [4] have emerged as particularly promising, offering comprehen￾sive assistance by leveraging anatomical information. These methods, rooted in an understanding of human anatomy, h… view at source ↗
Figure 2
Figure 2. Upper-limb exoskeleton structure with direct-drive joint modules for Joints 1 and 2, and cable-driven SEA actuation for Joints 3 to 5. TABLE I DYNAMIC PARAMETERS M(q) ∈ R5×5 Inertia matrix of robot C(q˙, q) ∈ R5×5 Matrix related to centripetal and Coriolis forces g(q) ∈ R5 Vector related to gravity K ∈ R3×3 Stiffness matrix q ∈ R5 Robot joint angles B ∈ R3×3 Inertia matrix of motor θ ∈ R3 Motor rotation angles τe ∈ … view at source ↗
Figure 3
Figure 3. The framework of ATP: A muscle controller with anatomical knowledge is rapidly trained on a GPU-accelerated musculoskeletal system by using RL. The real-time generated joint reference torque is refined online by incorporating safety constraints and an anomaly score with comfort information, serving as the desired human–robot interaction torque for the exoskeleton controller. Passivity of human–robot interaction is e… view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Kinematic mapping between the exoskeleton and musculoskeletal model for four-axis alignment, enabling delivered assistance to be transformed into the exoskeleton coordinate. where f(·) represents the forward kinematics, and J (·) is the Jacobian matrix that relates the…
Figure 5
Figure 5. Figure 5: Experimental setup of the upper-limb exoskeleton. IMUs are worn by the subject to measure limb angles for estimating the anatomical torque reference, while EMG sensors are mounted on different upper-limb muscles to evaluate region-specific muscle activity. measurement …
Figure 7
Figure 7. Figure 7: Training curve of the muscle controller. The tracking error across different tasks decreases as the controller continues to be jointly trained on all datasets. in which the wearer freely moves the arm to cover a broad range of motion patterns relevant to exoskeleton as…
Figure 8
Figure 8. Figure 8: Motion-tracking performance of the musculoskeletal system driven by the muscle controller. The controller is jointly trained on multiple trajectory datasets and then used to coordinate the activation of all muscles, enabling four independent joints to track a long-dura…
Figure 10
Figure 10. Figure 10: Interaction torque-tracking performance in simulation. The proposed passivity-based controller is implemented on a single-joint SEA in simulation and achieves accurate torque tracking when passivity is not violated. tracking targets for the muscle controller. To ensur…
Figure 9
Figure 9. Figure 9: Position tracking achieved by applying the muscle controller to unseen upper-limb movements in a real-world setup. The target positions are obtained by measuring joint angles in the exoskeleton coordinate using IMUs and subsequently mapping them to the musculoskeletal …
Figure 11
Figure 11. Figure 11: Simulation results of tracking performance and system power variations when depleted tanks are activated, represented by shaded areas: (a)–(b) torque tank activation; (c)–(d) observer tank activation [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Ablation study on the energy tank. Torque-tracking performance is compared with and without the energy tank under periodic torque commands. The net work injected into the joint in each cycle is also evaluated, where a positive value indicates that the system absorbs e…
Figure 14
Figure 14. Figure 14: Torque-tracking performance of the shoulder joint under tank state transition switching. The red shaded region represents tank depletion, while the green shaded region indicates that the wearer interacts with the exoskeleton to replenish tank, allowing the controller …
Figure 16
Figure 16. Figure 16: Torque refinement during the simulated collision. The red shaded region indicates the period during which the collision occurs, while the refined torque decreased accordingly to mitigate conflict. Approaching singularity [PITH_FULL_IMAGE:figures/full_fig_p013_16.png]
Figure 17
Figure 17. Figure 17: Torque refinement during motion toward a kinematic singularity. The red sector denotes the region approaching Jacobian singularity, while the red shaded area indicates the period during which the anomaly occurs. The refined torque increases accordingly, generating a c…
Figure 18
Figure 18. Figure 18: Performance of torque refinement compared with different torque profiles. The wearer performs a compound movement involving three joints, with activated tendons highlighted in dark red. Torque refinement improves the smoothness and safety of the delivered assistance w…
Figure 19
Figure 19. Figure 19: EMG responses of different muscles under various assistance conditions across multiple tasks. The proposed ATP controller assists the wearer during three static tasks and two dynamic tasks, reducing the EMG activity of the primary muscles engaged in each task. TABLE V…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 45 canonical work pages

  1. [1]

    (2026) Hypershell x exoskeleton

    Hypershell. (2026) Hypershell x exoskeleton. [Online]. Available: https://hypershell.tech/

    work page 2026

  2. [2]

    Design, development, and control of a hand/wrist exoskeleton for rehabilitation and training,

    M. Dragusanu, M. Z. Iqbal, T. L. Baldi, D. Prattichizzo, and M. Malvezzi, “Design, development, and control of a hand/wrist exoskeleton for rehabilitation and training,”IEEE Transactions on Robotics, vol. 38, no. 3, pp. 1472–1488, 2022

    work page 2022

  3. [3]

    Biomechanical design of the berkeley lower extremity exoskeleton (BLEEX),

    A. B. Zoss, H. Kazerooni, and A. Chu, “Biomechanical design of the berkeley lower extremity exoskeleton (BLEEX),”IEEE/ASME Transac- tions on mechatronics, vol. 11, no. 2, pp. 128–138, 2006

    work page 2006

  4. [4]

    Human- in-the-loop optimization of exoskeleton assistance via online simulation of metabolic cost,

    D. F. Gordon, C. McGreavy, A. Christou, and S. Vijayakumar, “Human- in-the-loop optimization of exoskeleton assistance via online simulation of metabolic cost,”IEEE Transactions on Robotics, vol. 38, no. 3, pp. 1410–1429, 2022

    work page 2022

  5. [5]

    Task-agnostic exoskeleton control via biological joint moment estimation,

    D. D. Molinaro, K. L. Scherpereel, E. B. Schonhaut, G. Evangelopoulos, M. K. Shepherd, and A. J. Young, “Task-agnostic exoskeleton control via biological joint moment estimation,”Nature, vol. 635, no. 8038, pp. 337–344, 2024

    work page 2024

  6. [6]

    Human-cooperative control design of a walking exoskeleton for body weight support,

    Z. Li, Z. Ren, K. Zhao, C. Deng, and Y . Feng, “Human-cooperative control design of a walking exoskeleton for body weight support,”IEEE Transactions on Industrial Informatics, vol. 16, no. 5, pp. 2985–2996, 2019

    work page 2019

  7. [7]

    High-accuracy early recognition of upper-limb motions for exoskeleton-assisted mirror reha- bilitation,

    H. Wang, Y . Yao, H. Lei, Y . Shi, and S. Pei, “High-accuracy early recognition of upper-limb motions for exoskeleton-assisted mirror reha- bilitation,”IEEE Robotics and Automation Letters, 2025

    work page 2025

  8. [8]

    A semantic- aware framework for safe and intent-integrative assistance in upper-limb exoskeletons,

    Y . Chen, S. Miao, C. Wu, J. Mu, B. Ouyang, and X. Li, “A semantic- aware framework for safe and intent-integrative assistance in upper-limb exoskeletons,”IEEE Robotics and Automation Letters, vol. 11, no. 5, pp. 5938–5945, 2026

    work page 2026

Show all 53 references
  1. [9]

    Mujoco playground,

    K. Zakka, B. Tabanpour, Q. Liao, M. Haiderbhai, S. Holt, J. Y . Luo, A. Allshire, E. Frey, K. Sreenath, L. A. Kahrs,et al., “Mujoco playground,”arXiv preprint arXiv:2502.08844, 2025

    2025 arXiv

  2. [10]

    Myosim: Fast and physiologically realistic mujoco models for mus- culoskeletal and exoskeletal studies,

    H. Wang, V . Caggiano, G. Durandau, M. Sartori, and V . Kumar, “Myosim: Fast and physiologically realistic mujoco models for mus- culoskeletal and exoskeletal studies,” in2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 8104–8111

    2022

  3. [11]

    mjlab: A lightweight framework for gpu-accelerated robot learning,

    K. Zakka, Q. Liao, B. Yi, L. L. Lay, K. Sreenath, and P. Abbeel, “mjlab: A lightweight framework for gpu-accelerated robot learning,” arXiv preprint arXiv:2601.22074, 2026

    2026

  4. [12]

    Emg-based model predictive control for physical human–robot interaction: Application for assist-as- needed control,

    T. Teramae, T. Noda, and J. Morimoto, “Emg-based model predictive control for physical human–robot interaction: Application for assist-as- needed control,”IEEE Robotics and Automation Letters, vol. 3, no. 1, pp. 210–217, 2017

    2017

  5. [13]

    The effects of incline level on optimized lower-limb exoskeleton assistance: A case series,

    P. W. Franks, G. M. Bryan, R. Reyes, M. P. O’Donovan, K. N. Gregorczyk, and S. H. Collins, “The effects of incline level on optimized lower-limb exoskeleton assistance: A case series,”IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 30, pp. 2494–2505, 2022

    2022

  6. [14]

    A low-profile hip exoskeleton for pathological gait assistance: design and pilot testing,

    S. S. P. A. Bishe, L. Liebelt, Y . Fang, and Z. F. Lerner, “A low-profile hip exoskeleton for pathological gait assistance: design and pilot testing,” in2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 5461–5466

    2022

  7. [15]

    Human-in-the-loop optimization of exoskeleton assistance during walking,

    J. Zhang, P. Fiers, K. A. Witte, R. W. Jackson, K. L. Poggensee, C. G. Atkeson, and S. H. Collins, “Human-in-the-loop optimization of exoskeleton assistance during walking,”Science, vol. 356, no. 6344, pp. 1280–1284, 2017

    2017

  8. [16]

    Automatically customizing a powered knee prosthesis with human in the loop using adaptive dynamic programming,

    Y . Wen, A. Brandt, J. Si, and H. H. Huang, “Automatically customizing a powered knee prosthesis with human in the loop using adaptive dynamic programming,” in2017 International Symposium on Wearable Robotics and Rehabilitation (WeRob). IEEE, 2017, pp. 1–2

    2017

  9. [17]

    Human-in-the- loop optimization of hip assistance with a soft exosuit during walking,

    Y . Ding, M. Kim, S. Kuindersma, and C. J. Walsh, “Human-in-the- loop optimization of hip assistance with a soft exosuit during walking,” Science Robotics, vol. 3, no. 15, p. eaar5438, 2018

    2018

  10. [18]

    Personal- izing exoskeleton assistance while walking in the real world,

    P. Slade, M. J. Kochenderfer, S. L. Delp, and S. H. Collins, “Personal- izing exoskeleton assistance while walking in the real world,”Nature, vol. 610, no. 7931, pp. 277–282, 2022

    2022

  11. [19]

    Learning to assist different wearers in multitasks: efficient and individualized human-in-the-loop adaptation framework for lower-limb exoskeleton,

    Y . Chen, S. Miao, G. Chen, J. Ye, C. Fu, B. Liang, S. Song, and X. Li, “Learning to assist different wearers in multitasks: efficient and individualized human-in-the-loop adaptation framework for lower-limb exoskeleton,”IEEE Transactions on Robotics, 2024

    2024

  12. [20]

    Opensim: Simulating musculoskeletal dynamics and neuromuscular control to study human and animal movement,

    A. Seth, J. L. Hicks, T. K. Uchida, A. Habib, C. L. Dembia, J. J. Dunne, C. F. Ong, M. S. DeMers, A. Rajagopal, M. Millard,et al., “Opensim: Simulating musculoskeletal dynamics and neuromuscular control to study human and animal movement,”PLoS computational biology, vol. 14, n...

    2018

  13. [21]

    Subject-independent, biological hip moment estimation during multi- modal overground ambulation using deep learning,

    D. D. Molinaro, I. Kang, J. Camargo, M. C. Gombolay, and A. J. Young, “Subject-independent, biological hip moment estimation during multi- modal overground ambulation using deep learning,”IEEE Transactions on Medical Robotics and Bionics, vol. 4, no. 1, pp. 219–229, 2022

    2022

  14. [22]

    Reinforce- ment learning methods for assistive and rehabilitation robotic systems: A survey,

    M. Sharifi, S. Tripathi, Y . Chen, Q. Zhang, and M. Tavakoli, “Reinforce- ment learning methods for assistive and rehabilitation robotic systems: A survey,”IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2025

    2025

  15. [23]

    End-to-end deep reinforcement learning for exoskeleton control,

    L. Rose, M. C. Bazzocchi, and G. Nejat, “End-to-end deep reinforcement learning for exoskeleton control,” in2020 IEEE international conference on systems, man, and cybernetics (SMC). IEEE, 2020, pp. 4294–4301

    2020

  16. [24]

    Experiment-free exoskeleton assistance via learning in simulation,

    S. Luo, M. Jiang, S. Zhang, J. Zhu, S. Yu, I. Dominguez Silva, T. Wang, E. Rouse, B. Zhou, H. Yuk,et al., “Experiment-free exoskeleton assistance via learning in simulation,”Nature, vol. 630, no. 8016, pp. 353–359, 2024

    2024

  17. [25]

    Robust walking control of a lower limb rehabilitation exoskeleton coupled with a musculoskeletal model via deep reinforcement learning,

    S. Luo, G. Androwis, S. Adamovich, E. Nunez, H. Su, and X. Zhou, “Robust walking control of a lower limb rehabilitation exoskeleton coupled with a musculoskeletal model via deep reinforcement learning,” Journal of neuroengineering and rehabilitation, vol. 20, no. 1, p. 34, 2023

    2023

  18. [26]

    Robust neural network controller for a lower limb exoskeleton with minimal sensor configuration: a deep reinforcement learning approach with policy distillation,

    S. Luo, K. Feng, G. Androwis, S. Adamovich, E. Nunez, H. Su, and X. Zhou, “Robust neural network controller for a lower limb exoskeleton with minimal sensor configuration: a deep reinforcement learning approach with policy distillation,”Journal of NeuroEngineering and Rehabili...

    2025

  19. [27]

    Predicting sit-to-stand motions with a deep reinforcement learning based controller under idealized exoskeleton assistance,

    N. Ratnakumar, K. Akba¸ s, R. Jones, Z. You, and X. Zhou, “Predicting sit-to-stand motions with a deep reinforcement learning based controller under idealized exoskeleton assistance,”Multibody System Dynamics, pp. 1–18, 2024

    2024

  20. [28]

    Control- ling fes of arm movements using physics-informed reinforcement learn- ing via co-kriging adjustment,

    N. Wannawas, C. Diaz-Pintado, J. Narayan, and A. A. Faisal, “Control- ling fes of arm movements using physics-informed reinforcement learn- ing via co-kriging adjustment,” in2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 5555–5560

    2024

  21. [29]

    Scalable muscle-actuated human simulation and control,

    S. Lee, M. Park, K. Lee, and J. Lee, “Scalable muscle-actuated human simulation and control,”ACM Transactions On Graphics (TOG), vol. 38, no. 4, pp. 1–13, 2019

    2019

  22. [30]

    Self model for embodied intelligence: Modeling full-body human musculoskeletal system and locomotion control with hierarchical low-dimensional representation,

    C. Zuo, K. He, J. Shao, and Y . Sui, “Self model for embodied intelligence: Modeling full-body human musculoskeletal system and locomotion control with hierarchical low-dimensional representation,” in2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2...

    2024

  23. [31]

    Towards embodied ai with musclemimic: Unlock- ing full-body musculoskeletal motor learning at scale,

    C. Li, C. Wang, B. Ziliotto, M. Simos, J. Kovecses, G. Durandau, and A. Mathis, “Towards embodied ai with musclemimic: Unlock- ing full-body musculoskeletal motor learning at scale,”arXiv preprint arXiv:2603.25544, 2026

    2026

  24. [32]

    Fuzzy adaptive controller of a wearable assistive upper limb exoskeleton using a disturbance observer,

    M. S. Amiri and R. Ramli, “Fuzzy adaptive controller of a wearable assistive upper limb exoskeleton using a disturbance observer,”IEEE Transactions on Human-Machine Systems, 2025

    2025

  25. [33]

    semg-based joint force control for an upper-limb power-assist exoskeleton robot,

    Z. Li, B. Wang, F. Sun, C. Yang, Q. Xie, and W. Zhang, “semg-based joint force control for an upper-limb power-assist exoskeleton robot,” IEEE journal of biomedical and health informatics, vol. 18, no. 3, pp. 1043–1050, 2013

    2013

  26. [34]

    Learning impedance control for robotic manipulators,

    C.-C. Cheah and D. Wang, “Learning impedance control for robotic manipulators,”IEEE Transactions on Robotics and Automation, vol. 14, no. 3, pp. 452–465, 1998

    1998

  27. [35]

    Intention-driven variable impedance control for physical human-robot interaction,

    Y . Huo, X. Li, X. Zhang, and D. Sun, “Intention-driven variable impedance control for physical human-robot interaction,” in2021 IEEE/ASME International Conference on Advanced Intelligent Mecha- tronics (AIM). IEEE, 2021, pp. 1220–1225

    2021

  28. [36]

    Safe and individualized motion planning for upper-limb exoskeleton robots using human demon- stration and interactive learning,

    Y . Chen, G. Chen, J. Ye, X. Qiu, and X. Li, “Safe and individualized motion planning for upper-limb exoskeleton robots using human demon- stration and interactive learning,” in2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 15 307–15 313

    2024

  29. [37]

    Estimating human joint moments unifies exoskeleton control, reducing user effort,

    D. D. Molinaro, I. Kang, and A. J. Young, “Estimating human joint moments unifies exoskeleton control, reducing user effort,”Science robotics, vol. 9, no. 88, p. eadi8852, 2024

    2024

  30. [38]

    On human-in-the-loop optimization of human–robot interaction,

    P. Slade, C. Atkeson, J. M. Donelan, H. Houdijk, K. A. Ingraham, M. Kim, K. Kong, K. L. Poggensee, R. Riener, M. Steinert,et al., “On human-in-the-loop optimization of human–robot interaction,”Nature, vol. 633, no. 8031, pp. 779–788, 2024

    2024

  31. [39]

    C. A. Desoer and M. Vidyasagar,Feedback systems: input-output properties. SIAM, 2009. 17

    2009

  32. [40]

    Passivity and stability of human–robot inter- action control for upper-limb rehabilitation robots,

    J. Zhang and C. C. Cheah, “Passivity and stability of human–robot inter- action control for upper-limb rehabilitation robots,”IEEE Transactions on Robotics, vol. 31, no. 2, pp. 233–245, 2015

    2015

  33. [41]

    Passivity preser- vation for variable impedance control of compliant robots,

    E. Spyrakos-Papastavridis, P. R. Childs, and J. S. Dai, “Passivity preser- vation for variable impedance control of compliant robots,”IEEE/ASME Transactions on Mechatronics, vol. 25, no. 5, pp. 2342–2353, 2019

    2019

  34. [42]

    Unified force-impedance control,

    S. Haddadin and E. Shahriari, “Unified force-impedance control,”The International Journal of Robotics Research, vol. 43, no. 13, pp. 2112– 2141, 2024

    2024

  35. [43]

    A unified passivity-based control framework for position, torque and impedance control of flexible joint robots,

    A. Albu-Schäffer, C. Ott, and G. Hirzinger, “A unified passivity-based control framework for position, torque and impedance control of flexible joint robots,”The International Journal of Robotics Research, vol. 26, no. 1, pp. 23–39, 2007

    2007

  36. [44]

    C. C. Cheah and X. Li,Task-space sensory feedback control of robot manipulators. Springer, 2015, vol. 73

    2015

  37. [45]

    Control theory of nonlinear mechanical systems,

    S. Arimoto, “Control theory of nonlinear mechanical systems,”A Passivity-based and Circuit-theoretic Approach, 1996

    1996

  38. [46]

    Upper- limb rehabilitation with a dual-mode individualized exoskeleton robot: A generative-model-based solution,

    Y . Chen, S. Miao, J. Ye, G. Chen, J. Cheng, K. Du, and X. Li, “Upper- limb rehabilitation with a dual-mode individualized exoskeleton robot: A generative-model-based solution,”The International Journal of Robotics Research, p. 02783649251333479, 2025

    2025

  39. [47]

    Kungfubot: Physics-based humanoid whole-body control for learning highly-dynamic skills,

    W. Xie, J. Han, J. Zheng, H. Li, X. Liu, J. Shi, W. Zhang, C. Bai, and X. Li, “Kungfubot: Physics-based humanoid whole-body control for learning highly-dynamic skills,”arXiv preprint arXiv:2506.12851, 2025

    2025 arXiv

  40. [48]

    Adaptive human–robot interaction control for robots driven by series elastic actuators,

    X. Li, Y . Pan, G. Chen, and H. Yu, “Adaptive human–robot interaction control for robots driven by series elastic actuators,”IEEE Transactions on Robotics, vol. 33, no. 1, pp. 169–182, 2016

    2016

  41. [49]

    Adaptive task-space regulation of rigid-link flexible-joint robots with uncertain kinematics,

    C. Liu, C. C. Cheah, and J.-J. E. Slotine, “Adaptive task-space regulation of rigid-link flexible-joint robots with uncertain kinematics,”Automatica, vol. 44, no. 7, pp. 1806–1814, 2008

    2008

  42. [50]

    Unified passivity-based cartesian force/impedance control for rigid and flexible joint robots via task- energy tanks,

    C. Schindlbeck and S. Haddadin, “Unified passivity-based cartesian force/impedance control for rigid and flexible joint robots via task- energy tanks,” in2015 IEEE international conference on robotics and automation (ICRA). IEEE, 2015, pp. 440–447

    2015

  43. [51]

    Valve-based virtual energy tanks: A framework to simultaneously passify controls and embed control objectives,

    E. Shahriari, L. Johannsmeier, and S. Haddadin, “Valve-based virtual energy tanks: A framework to simultaneously passify controls and embed control objectives,” in2018 Annual American Control Conference (ACC). IEEE, 2018, pp. 3634–3641

    2018

  44. [52]

    Ott,Cartesian impedance control of redundant and flexible-joint robots

    C. Ott,Cartesian impedance control of redundant and flexible-joint robots. Springer, 2008

    2008

  45. [53]

    Robot operating system 2: Design, architecture, and uses in the wild,

    S. Macenski, T. Foote, B. P. Gerkey, C. Lalancette, and W. Woodall, “Robot operating system 2: Design, architecture, and uses in the wild,”Science Robotics, vol. 7, 2022. [Online]. Available: https://api.semanticscholar.org/CorpusID:248736816 18

    2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.