pith. sign in

arxiv: 2605.02513 · v1 · submitted 2026-05-04 · 💻 cs.RO

Adaptive Gait Generation for Multi-Terrain Exoskeletons via Constrained Kernelized Movement Primitives

Edoardo Trombin , Miroljub Mihailovic , Matheus Henrique Ferreira Moura , Luca Tonin , Emanuele Menegatti , Stefano Tortora This is my paper

Pith reviewed 2026-05-08 17:49 UTC · model grok-4.3

classification 💻 cs.RO
keywords adaptive gait generationlower limb exoskeletonskernelized movement primitivesmulti-terrain adaptationconstrained optimizationRGB-D sensinghuman gait modeling
0
0 comments X

The pith

Lower limb exoskeletons can adapt human gait models learned from few demonstrations to slopes, stairs, and obstacles by treating adaptation as a linearly constrained optimization problem informed by RGB-D via-points.

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

The paper establishes a Kernelized Movement Primitives framework that learns probabilistic representations of human gait in joint and task space from a limited set of demonstrations. These models are then adapted in real time by solving a constrained optimization that incorporates environmental via-points extracted from an onboard RGB-D camera. The resulting trajectories remain kinematically feasible and physiologically consistent across indoor terrains. A sympathetic reader would care because current commercial exoskeletons are restricted to flat ground, limiting their practical use for everyday mobility assistance.

Core claim

By representing natural gait as probabilistic distributions in both joint and Cartesian space via Kernelized Movement Primitives and then enforcing linear constraints derived from camera-detected via-points, the method generates environment-aware walking trajectories that preserve human-like characteristics while satisfying the kinematic limits of a lower-limb exoskeleton on flat ground, slopes, stairs, and obstacle crossings.

What carries the argument

Kernelized Movement Primitives formulated as a linearly constrained optimization problem that incorporates RGB-D via-points to adapt learned gait distributions.

If this is right

  • Gait trajectories can be generated in real time for flat ground, slopes, stairs, and obstacle crossing without retraining.
  • The learned models ensure both kinematic feasibility and preservation of natural gait characteristics.
  • Validation occurs first in simulation across multiple scenarios and then in physical experiments on a commercial device.
  • The approach supports environment-aware planning using only onboard RGB-D sensing.

Where Pith is reading between the lines

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

  • The method could be extended to dynamic environments by updating via-points at higher frequency from continuous camera streams.
  • Similar constrained KMP adaptation might apply to upper-limb exoskeletons or collaborative robots where task-space constraints dominate.
  • If the linear-constraint formulation proves robust, it reduces the data requirements for deploying exoskeletons in new buildings or outdoor settings.

Load-bearing premise

Probabilistic gait models learned from a limited number of demonstrations on flat terrain can be adapted via linear constraints and RGB-D via-points to produce stable, kinematically feasible trajectories on unseen multi-terrain conditions.

What would settle it

Real-world trials on the commercial lower-limb exoskeleton that produce frequent balance loss, joint-limit violations, or visibly unnatural motion when ascending stairs or crossing obstacles would falsify the claim of reliable adaptation.

Figures

Figures reproduced from arXiv: 2605.02513 by Edoardo Trombin, Emanuele Menegatti, Luca Tonin, Matheus Henrique Ferreira Moura, Miroljub Mihailovic, Stefano Tortora.

Figure 1
Figure 1. Figure 1: Overview of the proposed solution for environment-adaptive gait generation. The view at source ↗
Figure 2
Figure 2. Figure 2: General kinematic model of the exoskeleton for the implementation view at source ↗
Figure 3
Figure 3. Figure 3: Examples of different terrains acquired with an RGB-D camera (left view at source ↗
Figure 4
Figure 4. Figure 4: Final exoskeleton configuration for local frame computation in different terrains. view at source ↗
Figure 5
Figure 5. Figure 5: Obstacle’s linear constraints identified from the ensemble of foot view at source ↗
Figure 6
Figure 6. Figure 6: Human demonstrations (blue) used to train the swing foot KMP view at source ↗
Figure 9
Figure 9. Figure 9: Swing foot trajectories for different obstacle sizes (support foot view at source ↗
Figure 8
Figure 8. Figure 8: Swing foot trajectories for different slope inclinations. view at source ↗
Figure 10
Figure 10. Figure 10: Results of the real experiments in different scenarios (time view at source ↗
Figure 11
Figure 11. Figure 11: Results of the adaptive gait generation experiments on the real exoskeleton across different terrains (flat-ground, curb ascending-descending, slope view at source ↗
Figure 12
Figure 12. Figure 12: Results of the adaptive gait generation experiments on the real exoskeleton when crossing different types of obstacles (cube, cylinder, rectangle, view at source ↗
read the original abstract

Lower limb exoskeletons (LLEs) present the potential to make motor-impaired individuals walk again. Their application in real-world environments is still limited by the lack of effective adaptive gait planning. Indeed, current exoskeletons are meant to walk only on a flat and even terrain. Generating environment-aware, physiologically consistent gait trajectories in real-time is an open challenge. To overcome this, we propose a novel Kernelized Movement Primitives (KMP)-based framework for adaptive gait generation (AGG) across multiple indoor terrains. The proposed approach learns a probabilistic representation of human gait in both the joint and task spaces from a limited number of human demonstrations, representing natural gait characteristics and ensuring kinematic feasibility. In addition, the learned trajectories are adapted using environmental information extracted from an onboard RGB-D camera by treating the AGG as a linearly constrained optimization problem with via-points. The proposed method has been thoroughly validated first in simulations for gait generation in different scenarios, such as flat-ground walking, slopes, stairs, and obstacles crossing. Finally, the effectiveness and robustness of the method have been demonstrated with experiments on a commercial LLE in real-world scenarios. The results obtained demonstrate the feasibility of an environment-aware gait planning system for a new generation of intelligent lower limb exoskeletons for assisting people with disabilities in their every-day life.

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.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a Kernelized Movement Primitives (KMP)-based Adaptive Gait Generation (AGG) framework for lower-limb exoskeletons that learns probabilistic gait models from a limited set of human demonstrations in joint and task space and then adapts them to unseen multi-terrain conditions (slopes, stairs, obstacles) by solving a linearly constrained optimization problem whose constraints are supplied by RGB-D camera via-points. The central claim is that the resulting trajectories remain kinematically feasible and physiologically consistent, with supporting evidence from simulation across four terrain classes and hardware trials on a commercial LLE.

Significance. If the kinematic adaptation reliably yields dynamically stable and physiologically natural gaits, the work would constitute a practical advance toward environment-aware exoskeleton control beyond the flat-terrain restriction of current commercial devices. The data-efficient probabilistic representation and the explicit use of onboard RGB-D sensing are attractive features; the dual simulation-plus-hardware validation further strengthens the practical relevance of the approach.

major comments (2)
  1. [§4] §4 (adaptation formulation): the AGG problem is posed as a linearly constrained KMP optimization whose only task-space constraints are RGB-D via-points. No explicit dynamic stability margins (ZMP, friction-cone, or minimum swing-foot clearance) are included in the optimizer. Because the central claim is that the adapted trajectories remain stable and physiologically consistent on slopes and stairs, the absence of these constraints is load-bearing and requires either an added dynamic layer or a quantitative demonstration that the purely kinematic solution never violates stability bounds.
  2. [Real-world experiments] Real-world experiments section: the manuscript states that effectiveness and robustness were demonstrated on a commercial LLE, yet reports neither quantitative error metrics (e.g., RMSE against reference gaits, foot-clearance statistics), statistical tests, nor baseline comparisons. Without these data it is impossible to evaluate whether the adaptation actually improves physiological consistency or merely produces feasible but potentially unstable trajectories.
minor comments (2)
  1. The number of human demonstrations and the precise terrain on which they were recorded are not stated explicitly; this information is needed to assess how well the learned prior generalizes to the four test terrains.
  2. Notation for the KMP kernel hyperparameters and the via-point weighting matrices should be introduced once in a dedicated table or appendix to improve readability of the optimization problem.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and insightful comments, which have helped us improve the clarity and rigor of the manuscript. We address each major comment below and have revised the paper accordingly.

read point-by-point responses

  1. Referee: [§4] §4 (adaptation formulation): the AGG problem is posed as a linearly constrained KMP optimization whose only task-space constraints are RGB-D via-points. No explicit dynamic stability margins (ZMP, friction-cone, or minimum swing-foot clearance) are included in the optimizer. Because the central claim is that the adapted trajectories remain stable and physiologically consistent on slopes and stairs, the absence of these constraints is load-bearing and requires either an added dynamic layer or a quantitative demonstration that the purely kinematic solution never violates stability bounds.

    Authors: We appreciate the referee's point that dynamic stability is central to the claims. Our formulation prioritizes a kinematic, data-efficient approach for real-time onboard execution on embedded hardware, where full dynamic optimization would be prohibitive. However, we agree that explicit verification is needed. In the revised manuscript we have added a post-adaptation stability analysis in §4, computing ZMP trajectories (projected onto the support polygon) and minimum swing-foot clearance for all simulated and hardware trials. The results confirm that ZMP remains inside the base of support and clearance exceeds 5 cm on average across slopes, stairs, and obstacles, providing quantitative evidence that the kinematic solutions satisfy the cited stability bounds without requiring a dynamic layer in the optimizer. revision: yes

  2. Referee: [Real-world experiments] Real-world experiments section: the manuscript states that effectiveness and robustness were demonstrated on a commercial LLE, yet reports neither quantitative error metrics (e.g., RMSE against reference gaits, foot-clearance statistics), statistical tests, nor baseline comparisons. Without these data it is impossible to evaluate whether the adaptation actually improves physiological consistency or merely produces feasible but potentially unstable trajectories.

    Authors: We acknowledge that the original real-world section lacked sufficient quantitative detail. The revised manuscript now reports RMSE for both joint-space and task-space trajectories against reference human gaits, mean and minimum foot-clearance statistics, and success rates over repeated trials. We have also added a baseline comparison against unconstrained KMP and performed paired statistical tests (Wilcoxon signed-rank) demonstrating significant improvements in tracking accuracy and clearance. These metrics and tests are presented in updated tables and figures in the real-world experiments section, allowing direct evaluation of physiological consistency and stability. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation relies on external demonstrations and independent optimization

full rationale

The paper's core chain learns a probabilistic KMP model from a limited set of external human gait demonstrations, then adapts the resulting distribution by solving a linearly constrained optimization problem whose via-points are supplied by an independent RGB-D sensor. Neither step reduces to a self-definition, a fitted parameter renamed as prediction, or a load-bearing self-citation; the adaptation equations treat the learned mean and covariance as fixed inputs and add external linear constraints. Validation proceeds through separate simulation trials and hardware experiments on unseen terrains rather than any closed loop that presupposes the target trajectories. The method therefore contains independent empirical content and does not collapse to its own inputs by construction.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

Review based on abstract only; full paper likely contains additional implementation details on kernel selection and constraint formulation.

free parameters (2)
  • KMP kernel hyperparameters
    Parameters controlling the probabilistic representation of gait trajectories are typically fitted or tuned from demonstration data.
  • Via-point constraint weights
    Weights balancing task-space constraints from camera data against learned priors are chosen to ensure feasibility.
axioms (2)
  • domain assumption Limited human gait demonstrations capture sufficient natural characteristics and kinematic feasibility for generalization.
    Invoked in the learning step from demonstrations to represent joint and task space trajectories.
  • domain assumption RGB-D camera data can be reliably converted into accurate via-points for real-time optimization.
    Assumed when treating adaptation as a linearly constrained optimization problem.

pith-pipeline@v0.9.0 · 5563 in / 1476 out tokens · 36342 ms · 2026-05-08T17:49:03.693777+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

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

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

70 extracted references · 70 canonical work pages

  1. [1]

    Overview: Types of lower limb exoskeletons,

    D. S. Pamungkas, W. Caesarendra, H. Soebakti, R. Analia, and S. Su- santo, “Overview: Types of lower limb exoskeletons,”Electronics, vol. 8, no. 11, pp. 1283–1295, 2019

    work page 2019

  2. [2]

    Systematic Review of Exoskeletons towards a General Categorization Model Proposal,

    J. A. de la Tejera, R. Bustamante-Bello, R. A. Ramirez-Mendoza, and J. Izquierdo-Reyes, “Systematic Review of Exoskeletons towards a General Categorization Model Proposal,”Applied Sciences, vol. 11, no. 1, pp. 76–101, Jan. 2021

    work page 2021

  3. [3]

    Improvement of quality of life after 2- month exoskeleton training in patients with chronic spinal cord injury,

    I. J. van Nes, R. B. van Dijsseldonk, F. H. van Herpen, H. Rijken, A. C. Geurts, and N. L. Keijsers, “Improvement of quality of life after 2- month exoskeleton training in patients with chronic spinal cord injury,” The Journal of Spinal Cord Medicine, vol. 47, no. 3, pp. 354–360, 2024

    work page 2024

  4. [4]

    Mobile Exoskeleton for Spinal Cord Injury: Development and Testing,

    K. A. Strausser, T. A. Swift, A. B. Zoss, H. Kazerooni, and B. C. Bennett, “Mobile Exoskeleton for Spinal Cord Injury: Development and Testing,” inDynamic Systems and Control Conference, May 2012, pp. 419–425

    work page 2012

  5. [5]

    The ReWalk Powered Exoskeleton to Restore Ambulatory Function to Individuals with Thoracic-Level Motor-Complete Spinal Cord Injury,

    A. Esquenazi, M. Talaty, A. Packel, and M. Saulino, “The ReWalk Powered Exoskeleton to Restore Ambulatory Function to Individuals with Thoracic-Level Motor-Complete Spinal Cord Injury,”American Journal of Physical Medicine & Rehabilitation, vol. 91, no. 11, pp. 911–921, Nov. 2012

    work page 2012

  6. [6]

    Advancements in sensor technologies and control strategies for lower- limb rehabilitation exoskeletons: A comprehensive review,

    Y . Yao, D. Shao, M. Tarabini, S. A. Moezi, K. Li, and P. Saccomandi, “Advancements in sensor technologies and control strategies for lower- limb rehabilitation exoskeletons: A comprehensive review,”Microma- chines, vol. 15, no. 4, p. 489, 2024

    work page 2024

  7. [7]

    Review of Vision- Based Environmental Perception for Lower-Limb Exoskeleton Robots,

    C. Wang, Z. Pei, Y . Fan, S. Qiu, and Z. Tang, “Review of Vision- Based Environmental Perception for Lower-Limb Exoskeleton Robots,” Biomimetics, vol. 9, no. 4, Apr. 2024

    work page 2024

  8. [8]

    Vision-Assisted Autonomous Lower-Limb Exoskeleton Robot,

    D.-X. Liu, J. Xu, C. Chen, X. Long, D. Tao, and X. Wu, “Vision-Assisted Autonomous Lower-Limb Exoskeleton Robot,”IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 51, no. 6, pp. 3759–3770, June 2021

    work page 2021

  9. [9]

    A fast parameterized gait planning method for a lower-limb exoskeleton robot,

    H. Ren, W. Shang, N. Li, and X. Wu, “A fast parameterized gait planning method for a lower-limb exoskeleton robot,”International Journal of Advanced Robotic Systems, vol. 17, no. 1, Jan. 2020

    work page 2020

  10. [10]

    On-line Dynamic Gait Generation Model for Wearable Robot with User’s Motion Inten- tion,

    H. Ren, D.-X. Liu, N. Li, Y . He, Z. Yan, and X. Wu, “On-line Dynamic Gait Generation Model for Wearable Robot with User’s Motion Inten- tion,” inIEEE International Conference on Information and Automation (ICIA), Aug. 2018, pp. 347–352

    work page 2018

  11. [11]

    Online gait generator for lower limb exoskeleton robots: Suitable for level ground, slopes, stairs, and obstacle avoidance,

    H. Mohamad and S. Ozgoli, “Online gait generator for lower limb exoskeleton robots: Suitable for level ground, slopes, stairs, and obstacle avoidance,”Robotics and Autonomous Systems, vol. 160, p. 104319, 2023

    work page 2023

  12. [12]

    Minimum-Time and Minimum-Jerk Gait Planning in Joint Space for Assistive Lower Limb Exoskeleton,

    H. Mohamad, S. Ozgoli, and F. Motawej, “Minimum-Time and Minimum-Jerk Gait Planning in Joint Space for Assistive Lower Limb Exoskeleton,”Journal of Bionic Engineering, vol. 20, no. 5, pp. 2164– 2178, Mar. 2023

    work page 2023

  13. [13]

    Real-time walking pattern generation for a lower limb exoskeleton, implemented on the Exoped robot,

    J. Kazemi and S. Ozgoli, “Real-time walking pattern generation for a lower limb exoskeleton, implemented on the Exoped robot,”Robotics and Autonomous Systems, vol. 116, pp. 1–23, June 2019

    work page 2019

  14. [14]

    Optimization of Joint Space Trajectories for Assistive Lower Limb Exoskeleton Robots: Real-Time and Flexible Gait Patterns,

    H. Mohamad, S. Ozgoli, and J. Kazemi, “Optimization of Joint Space Trajectories for Assistive Lower Limb Exoskeleton Robots: Real-Time and Flexible Gait Patterns,”Journal of Intelligent & Robotic Systems, vol. 110, no. 3, p. 122, Aug. 2024

    work page 2024

  15. [15]

    Environment- adaptive gait planning for obstacle avoidance in lower-limb robotic exoskeletons,

    E. Trombin, S. Tortora, E. Menegatti, and L. Tonin, “Environment- adaptive gait planning for obstacle avoidance in lower-limb robotic exoskeletons,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2024, pp. 13 640–13 647

    work page 2024

  16. [16]

    Human-in-the-Loop Trajectory Optimization Based on sEMG Biofeedback for Lower-Limb Exoskeleton,

    L.-L. Li, Y .-P. Zhang, G.-Z. Cao, and W.-Z. Li, “Human-in-the-Loop Trajectory Optimization Based on sEMG Biofeedback for Lower-Limb Exoskeleton,”MDPI Sensors, vol. 24, no. 17, Jan. 2024

    work page 2024

  17. [17]

    Vision Assisted Control of Lower Extremity Exoskeleton for Obstacle Avoidance With Dynamic Constraint Based Piecewise Nonlinear MPC,

    Y . Hua, H. Zhang, Y . Li, J. Zhao, and Y . Zhu, “Vision Assisted Control of Lower Extremity Exoskeleton for Obstacle Avoidance With Dynamic Constraint Based Piecewise Nonlinear MPC,”IEEE Robotics and Automation Letters, vol. 7, no. 4, pp. 12 267–12 274, Oct. 2022

    work page 2022

  18. [18]

    Research on the Motion Control Strategy of a Lower-Limb Exoskeleton Rehabilitation Robot Using the Twin Delayed Deep Deterministic Policy Gradient Algorithm,

    Y . Guo, M. He, X. Tong, M. Zhang, and L. Huang, “Research on the Motion Control Strategy of a Lower-Limb Exoskeleton Rehabilitation Robot Using the Twin Delayed Deep Deterministic Policy Gradient Algorithm,”MDPI Sensors, vol. 24, no. 18, Jan. 2024

    work page 2024

  19. [19]

    Environment-Adaptive Gait Planning through Reinforcement Learning for Lower-Limb Exoskeletons,

    E. Trombin, F. Crisci, L. Tonin, E. Menegatti, and S. Tortora, “Environment-Adaptive Gait Planning through Reinforcement Learning for Lower-Limb Exoskeletons,” in2025 IEEE International Conference on Simulation, Modeling, and Programming for Autonomous Robots (SIMPAR), Apr. 2025, pp. 1–6

    work page 2025

  20. [20]

    Dynamic Movement Primitives -A Framework for Motor Control in Humans and Humanoid Robotics,

    S. Schaal, “Dynamic Movement Primitives -A Framework for Motor Control in Humans and Humanoid Robotics,” inAdaptive Motion of Animals and Machines, H. Kimura, K. Tsuchiya, A. Ishiguro, and H. Witte, Eds. Tokyo: Springer-Verlag, 2006, pp. 261–280

    work page 2006

  21. [21]

    Dynamical Movement Primitives: Learning Attractor Models for Motor Behaviors,

    A. J. Ijspeert, J. Nakanishi, H. Hoffmann, P. Pastor, and S. Schaal, “Dynamical Movement Primitives: Learning Attractor Models for Motor Behaviors,”Neural Computation, vol. 25, no. 2, pp. 328–373, Feb. 2013

    work page 2013

  22. [22]

    Adap- tive Gait Planning with Dynamic Movement Primitives for Walking Assistance Lower Exoskeleton in Uphill Slopes,

    R. Huang, Q. Wu, J. Qiu, H. Cheng, Q. Chen, and Z. Peng, “Adap- tive Gait Planning with Dynamic Movement Primitives for Walking Assistance Lower Exoskeleton in Uphill Slopes,”Sensors and Materials, vol. 32, no. 4, Apr. 2020

    work page 2020

  23. [23]

    Learning and planning of stair ascent for lower-limb exoskeleton systems,

    Q. Chen, H. Cheng, R. Huang, J. Qiu, and X. Chen, “Learning and planning of stair ascent for lower-limb exoskeleton systems,”Industrial Robot: the international journal of robotics research and application, vol. 46, no. 3, pp. 421–430, Jan. 2019

    work page 2019

  24. [24]

    Kernelized movement primitives,

    Y . Huang, L. Rozo, J. Silv ´erio, and D. G. Caldwell, “Kernelized movement primitives,”The International Journal of Robotics Research, vol. 38, no. 7, pp. 833–852, 2019

    work page 2019

  25. [25]

    A variable impedance skill learning algorithm based on kernelized movement primitives,

    A. Liu, S. Zhan, Z. Jin, and W.-A. Zhang, “A variable impedance skill learning algorithm based on kernelized movement primitives,”IEEE Transactions on Industrial Electronics, vol. 71, no. 1, pp. 870–879, 2023

    work page 2023

  26. [26]

    Collaborative robot trajectory tracking control based on ds-kmp algorithm,

    Y . Liu, Y . Cao, and C. Jiang, “Collaborative robot trajectory tracking control based on ds-kmp algorithm,”International Journal of Intelligent Robotics and Applications, pp. 1–13, 2025

    work page 2025

  27. [27]

    Reactive whole- body locomotion-integrated manipulation based on combined learning and optimization,

    J. Zhao, T. Teng, E. De Momi, and A. Ajoudani, “Reactive whole- body locomotion-integrated manipulation based on combined learning and optimization,”Machine Intelligence Research, pp. 1–14, 2025

    work page 2025

  28. [28]

    Leveraging kernelized synergies on shared subspace for precision grasping and dexterous manipulation,

    S. Katyara, F. Ficuciello, D. G. Caldwell, B. Siciliano, and F. Chen, “Leveraging kernelized synergies on shared subspace for precision grasping and dexterous manipulation,”IEEE Transactions on Cognitive and Developmental Systems, vol. 15, no. 4, pp. 2064–2076, 2021

    work page 2064

  29. [29]

    Humanoid grasping with multi-finger dexterous hands based on dm-kmp,

    J. Wang and K. Li, “Humanoid grasping with multi-finger dexterous hands based on dm-kmp,” in2024 WRC Symposium on Advanced Robotics and Automation (WRC SARA). IEEE, 2024, pp. 34–39

    work page 2024

  30. [30]

    Dimensionality reduction for probabilistic movement primitives,

    A. Colom ´e, G. Neumann, J. Peters, and C. Torras, “Dimensionality reduction for probabilistic movement primitives,” in2014 IEEE-RAS International Conference on Humanoid Robots. IEEE, 2014, pp. 794– 800

    work page 2014

  31. [31]

    Human-like trajectory planning based on postural synergistic kernelized movement primitives for robot-assisted rehabilitation,

    Z. Liu, Q. Ai, H. Liu, W. Meng, and Q. Liu, “Human-like trajectory planning based on postural synergistic kernelized movement primitives for robot-assisted rehabilitation,”IEEE Transactions on Human-Machine Systems, vol. 54, no. 2, pp. 152–161, 2024

    work page 2024

  32. [32]

    Dynamic Balance Gait for Walking Assistance Exoskeleton,

    Q. Chen, H. Cheng, C. Yue, R. Huang, and H. Guo, “Dynamic Balance Gait for Walking Assistance Exoskeleton,”Applied Bionics and Biomechanics, vol. 2018, no. 1, July 2018

    work page 2018

  33. [33]

    Hierarchical learn- ing control with physical human-exoskeleton interaction,

    R. Huang, H. Cheng, H. Guo, X. Lin, and J. Zhang, “Hierarchical learn- ing control with physical human-exoskeleton interaction,”Information Sciences, vol. 432, pp. 584–595, Mar. 2018

    work page 2018

  34. [34]

    Trajectory Generation and Control of a Lower Limb Exoskeleton for Gait Assistance,

    L. Luo, M. J. Foo, M. Ramanathan, J. K. Er, C. H. Chiam, L. Li, W. Y . Yau, and W. T. Ang, “Trajectory Generation and Control of a Lower Limb Exoskeleton for Gait Assistance,”Journal of Intelligent & Robotic Systems, vol. 106, no. 64, Nov. 2022

    work page 2022

  35. [35]

    Study of Lower Limb Exoskeleton Stair Movement Based on Multicoupled Continuous Dynamic Primitive Gait Learning Strategy,

    P. Zhang, J. Zhang, and J. Jia, “Study of Lower Limb Exoskeleton Stair Movement Based on Multicoupled Continuous Dynamic Primitive Gait Learning Strategy,”IEEE Sensors Journal, vol. 24, no. 2, pp. 2009– 2019, Jan. 2024

    work page 2009

  36. [36]

    Gait Planning with Dy- namic Movement Primitives for Lower Limb Exoskeleton Walking Up Stairs,

    W. Ma, H. Cheng, R. Huang, and Q. Chen, “Gait Planning with Dy- namic Movement Primitives for Lower Limb Exoskeleton Walking Up Stairs,” inIEEE International Conference on Robotics and Biomimetics (ROBIO), Dec. 2018, pp. 703–708

    work page 2018

  37. [37]

    Slope Gradient Adaptive Gait Planning for Walking Assistance Lower Limb Exoskele- tons,

    C. Zou, R. Huang, J. Qiu, Q. Chen, and H. Cheng, “Slope Gradient Adaptive Gait Planning for Walking Assistance Lower Limb Exoskele- tons,”IEEE Transactions on Automation Science and Engineering, vol. 18, no. 2, pp. 405–413, Apr. 2021

    work page 2021

  38. [38]

    Adaptive Gait Planning for Walking Assistance Lower Limb Exoskeletons in Slope Scenarios,

    C. Zou, R. Huang, H. Cheng, Q. Chen, and J. Qiu, “Adaptive Gait Planning for Walking Assistance Lower Limb Exoskeletons in Slope Scenarios,” in2019 International Conference on Robotics and Automa- tion (ICRA), May 2019, pp. 5083–5089. 16

    work page 2019

  39. [39]

    DMP-Based Motion Gener- ation for a Walking Exoskeleton Robot Using Divergent Component of Motion,

    D. Xu, P. Huang, Z. Li, and Y . Feng, “DMP-Based Motion Gener- ation for a Walking Exoskeleton Robot Using Divergent Component of Motion,” inInternational Conference on Advanced Robotics and Mechatronics (ICARM), July 2022, pp. 232–237

    work page 2022

  40. [40]

    Motion generation for walking exoskeleton robot using multiple dynamic movement primitives sequences combined with reinforcement learning,

    P. Zhang and J. Zhang, “Motion generation for walking exoskeleton robot using multiple dynamic movement primitives sequences combined with reinforcement learning,”Robotica, vol. 40, no. 8, pp. 2732–2747, Aug. 2022

    work page 2022

  41. [41]

    Learning Stylistic Dynamic Movement Primitives from multiple demonstrations,

    T. Matsubara, S.-H. Hyon, and J. Morimoto, “Learning Stylistic Dynamic Movement Primitives from multiple demonstrations,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, Oct. 2010, pp. 1277–1283

    work page 2010

  42. [42]

    Dynamic Movement Primitives based Parametric Gait Model for Lower Limb Exoskeleton,

    W. Ma, R. Huang, Q. Chen, G. Song, and C. Li, “Dynamic Movement Primitives based Parametric Gait Model for Lower Limb Exoskeleton,” in39th Chinese Control Conference (CCC), July 2020, pp. 3857–3862

    work page 2020

  43. [43]

    Probabilis- tic movement primitives,

    A. Paraschos, C. Daniel, J. R. Peters, and G. Neumann, “Probabilis- tic movement primitives,”Advances in neural information processing systems, vol. 26, 2013

    work page 2013

  44. [44]

    Probabilistic movement primitive based motion learning for a lower limb exoskeleton with black- box optimization,

    J. Wang, Y . Gao, D. Wu, and W. Dong, “Probabilistic movement primitive based motion learning for a lower limb exoskeleton with black- box optimization,”Frontiers of Information Technology & Electronic Engineering, vol. 24, no. 1, pp. 104–116, Jan. 2023

    work page 2023

  45. [45]

    Learning Gait Models With Varying Walking Speeds,

    C. Zou, R. Huang, H. Cheng, and J. Qiu, “Learning Gait Models With Varying Walking Speeds,”IEEE Robotics and Automation Letters, vol. 6, no. 1, pp. 183–190, Jan. 2021

    work page 2021

  46. [46]

    Terrain- Adaptive Gait Planning for Lower Limb Walking Assistance Exoskele- ton Robots,

    C. Zou, Z. Peng, K. Shi, F. Mu, R. Huang, and H. Cheng, “Terrain- Adaptive Gait Planning for Lower Limb Walking Assistance Exoskele- ton Robots,” in42nd Chinese Control Conference (CCC), July 2023, pp. 4773–4779

    work page 2023

  47. [47]

    Adaptive human-like gait planning for stair climbing of lower limb exoskeleton robots,

    C. Yang, X. Zhang, C. Zou, W. Liang, Z. Huang, R. Huang, Y . Wang, and H. Cheng, “Adaptive human-like gait planning for stair climbing of lower limb exoskeleton robots,” inInternational Conference on Intelligent Robotics and Applications. Springer, 2024, pp. 202–221

    work page 2024

  48. [48]

    A scoping review on lower limb exoskeleton actuation’s description and characteristics,

    F. Bettella, S. Tortora, E. Menegatti, N. Petrone, and A. Del Felice, “A scoping review on lower limb exoskeleton actuation’s description and characteristics,”Robotica, pp. 1–18, 2025

    work page 2025

  49. [49]

    Mlesac: A new robust estimator with application to estimating image geometry,

    P. Torr and A. Zisserman, “Mlesac: A new robust estimator with application to estimating image geometry,”Computer Vision and Image Understanding, vol. 78, no. 1, pp. 138–156, 2000. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S1077314299908329

    work page 2000

  50. [50]

    The spinal control of locomotion and step-to-step variability in left-right symmetry from slow to moderate speeds,

    C. Dambreville, A. Labarre, Y . Thibaudier, M.-F. Hurteau, and A. Frigon, “The spinal control of locomotion and step-to-step variability in left-right symmetry from slow to moderate speeds,”Journal of neurophysiology, vol. 114, no. 2, pp. 1119–1128, 2015

    work page 2015

  51. [51]

    Gmm-based single-joint angle estimation using emg signals,

    S. Michieletto, L. Tonin, M. Antonello, R. Bortoletto, F. Spolaor, E. Pagello, and E. Menegatti, “Gmm-based single-joint angle estimation using emg signals,” inIntelligent Autonomous Systems 13: Proceedings of the 13th International Conference IAS-13. Springer, 2015, pp. 1173– 1184

    work page 2015

  52. [52]

    A Linearly Constrained Nonparametric Framework for Imitation Learning,

    Y . Huang and D. G. Caldwell, “A Linearly Constrained Nonparametric Framework for Imitation Learning,” in2020 IEEE International Con- ference on Robotics and Automation (ICRA). IEEE, May 2020, pp. 4400–4406

    work page 2020

  53. [53]

    Comprehensive human locomotion and electromyography dataset: Gait120,

    J. Boo, D. Seo, M. Kim, and S. Koo, “Comprehensive human locomotion and electromyography dataset: Gait120,”Scientific Data, vol. 12, no. 1, June 2025

    work page 2025

  54. [54]

    A biomechanics dataset of healthy human walking at various speeds, step lengths and step widths,

    T. J. Van Der Zee, E. M. Mundinger, and A. D. Kuo, “A biomechanics dataset of healthy human walking at various speeds, step lengths and step widths,”Scientific Data, vol. 9, no. 1, p. 704, Nov. 2022

    work page 2022

  55. [55]

    Outdoor walking speeds of apparently healthy adults: A systematic review and meta-analysis,

    E. M. Murtagh, J. L. Mair, E. Aguiar, C. Tudor-Locke, and M. H. Murphy, “Outdoor walking speeds of apparently healthy adults: A systematic review and meta-analysis,”Sports Medicine, vol. 51, pp. 125– 141, 2021

    work page 2021

  56. [56]

    A guide to appropriate use of correlation coefficient in medical research,

    M. M. Mukaka, “A guide to appropriate use of correlation coefficient in medical research,”Malawi medical journal, vol. 24, no. 3, pp. 69–71, 2012

    work page 2012

  57. [57]

    A compre- hensive, open-source dataset of lower limb biomechanics in multiple conditions of stairs, ramps, and level-ground ambulation and transitions,

    J. Camargo, A. Ramanathan, W. Flanagan, and A. Young, “A compre- hensive, open-source dataset of lower limb biomechanics in multiple conditions of stairs, ramps, and level-ground ambulation and transitions,” Journal of Biomechanics, p. 110320, 2021

    work page 2021

  58. [58]

    Ros: an open-source robot operating system,

    M. Quigley, K. Conley, B. Gerkey, J. Faust, T. Foote, J. Leibs, R. Wheeler, A. Y . Ng,et al., “Ros: an open-source robot operating system,” inICRA workshop on open source software, vol. 3, no. 3.2. Kobe, 2009, p. 5

    work page 2009

  59. [59]

    An adaptive stair- ascending gait generation approach based on depth camera for lower limb exoskeleton,

    X. Zhao, W.-H. Chen, B. Li, X. Wu, and J. Wang, “An adaptive stair- ascending gait generation approach based on depth camera for lower limb exoskeleton,”Review of Scientific Instruments, vol. 90, no. 12, 2019

    work page 2019

  60. [60]

    Terrain slope parameter recognition for exoskeleton robot in urban multi-terrain environments,

    R. Guo, W. Li, Y . He, T. Zeng, B. Li, G. Song, and J. Qiu, “Terrain slope parameter recognition for exoskeleton robot in urban multi-terrain environments,”Complex & Intelligent Systems, vol. 10, no. 2, pp. 3107– 3118, 2024

    work page 2024

  61. [61]

    Vision-based stair environment perception for lower-limb ex- oskeletons using deep semantic segmentation,

    J. Zhang, “Vision-based stair environment perception for lower-limb ex- oskeletons using deep semantic segmentation,” in2025 6th International Conference on Computer Vision, Image and Deep Learning (CVIDL). IEEE, 2025, pp. 1261–1264

    work page 2025

  62. [62]

    Knee exoskeletons for gait rehabilitation and human performance augmentation: A state-of- the-art,

    B. Chen, B. Zi, Z. Wang, L. Qin, and W.-H. Liao, “Knee exoskeletons for gait rehabilitation and human performance augmentation: A state-of- the-art,”Mechanism and Machine Theory, vol. 134, pp. 499–511, 2019

    work page 2019

  63. [63]

    Estimation of quasi- stiffness of the human hip in the stance phase of walking,

    K. Shamaei, G. S. Sawicki, and A. M. Dollar, “Estimation of quasi- stiffness of the human hip in the stance phase of walking,”PloS one, vol. 8, no. 12, p. e81841, 2013

    work page 2013

  64. [64]

    Adaptive gait planning with dynamic movement primitives for walking assistance lower exoskeleton in uphill slopes

    R. Huang, Q. Wu, J. Qiu, H. Cheng, Q. Chen, and Z. Peng, “Adaptive gait planning with dynamic movement primitives for walking assistance lower exoskeleton in uphill slopes.”Sensors & Materials, vol. 32, 2020

    work page 2020

  65. [65]

    Online gait generator for lower limb exoskeleton robots: Suitable for level ground, slopes, stairs, and obstacle avoidance,

    H. Mohamad and S. Ozgoli, “Online gait generator for lower limb exoskeleton robots: Suitable for level ground, slopes, stairs, and obstacle avoidance,”Robotics and Autonomous Systems, vol. 160, p. 104319, Feb. 2023

    work page 2023

  66. [66]

    Ergonomic dual four- bar linkage knee exoskeleton for stair ascent assistance,

    S. Kittisares, T. Ide, H. Nabae, and K. Suzumori, “Ergonomic dual four- bar linkage knee exoskeleton for stair ascent assistance,”Frontiers in Robotics and AI, vol. 10, p. 1285520, 2023

    work page 2023

  67. [67]

    An adaptive gait planner for a lower-limb exoskeleton ascending staircases of unknown geometry,

    M. Raineri and C. G. Lo Bianco, “An adaptive gait planner for a lower-limb exoskeleton ascending staircases of unknown geometry,” in 2024 IEEE 20th International Conference on Automation Science and Engineering (CASE), Aug. 2024, pp. 2714–2719

    work page 2024

  68. [68]

    Gait speed using powered robotic exoskeletons after spinal cord injury: a systematic review and correlational study,

    D. R. Louie, J. J. Eng, T. Lam, and S. C. I. R. E. S. R. Team, “Gait speed using powered robotic exoskeletons after spinal cord injury: a systematic review and correlational study,”Journal of neuroengineering and rehabilitation, vol. 12, no. 1, p. 82, 2015

    work page 2015

  69. [69]

    Unsupervised sim- to-real adaptation for environmental recognition in assistive walking,

    C. Chen, K. Zhang, Y . Leng, X. Chen, and C. Fu, “Unsupervised sim- to-real adaptation for environmental recognition in assistive walking,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 30, pp. 1350–1360, 2022

    work page 2022

  70. [70]

    Exosense: A Vision-Based Scene Under- standing System For Exoskeletons,

    J. Wang, M. Mattamala, C. Kassab, G. Burger, F. Elnecave, L. Zhang, M. Petriaux, and M. Fallon, “Exosense: A Vision-Based Scene Under- standing System For Exoskeletons,” Nov. 2024

    work page 2024