Pith. sign in

REVIEW 4 major objections 7 minor 34 references

Context-aware collaborative pushing of heavy objects using skeleton-based intention prediction

T0 review · 4 major / 7 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 robot that reads the human skeleton can significantly reduce human effort when two agents slide a heavy object across a frictional surface, even though the robot cannot measure the human's force directly.

desk verdict A clean proof-of-concept that skeleton-based intention prediction can drive assistive pushing, but the experiments compare the whole system only to a no-robot baseline, so the headline claim about the predictor itself is stronger than the data support. read the letter →

arxiv 2505.10239 v1 pith:SMM27SPV submitted 2025-05-15 cs.RO

Gokhan Solak , Gustavo J. G. Lahr , Idil Ozdamar , Arash Ajoudani This is my paper · ORCID
classification cs.RO
keywords physicalhuman-robotinteractionintentionpredictionskeleton-basedactionrecognitiondirectedgraphneuralnetworkcollaborativepushingfrictioncompensationforce-freepHRIhumaneffortreduction
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 tries to establish that human motion intention — push, pull, or idle — can be read from skeleton posture alone, and that a robot using this prediction to apply a friction-compensating force can significantly reduce human effort in collaborative sliding of heavy objects. This matters because the standard physical human-robot interaction approach, admittance control driven by force feedback, is unusable when the object is not instrumented and friction masks the human force. The authors train a Directed Graph Neural Network on skeleton data, predict intention before the human applies significant force, and command the robot with force $f_d = i_h f_{com}$. Real-robot experiments report reduced mean and cumulative human force across different object weights, speeds, and two participants. The paper presents this as a proof-of-concept for a force-free, posture-based alternative to force sensing in physical human-robot interaction.

What carries the argument

The central machinery is a Directed Graph Neural Network (DGNN), an action-recognition architecture that represents the skeleton as a directed graph of joints and bones and updates vertex and edge attributes over multiple layers, with temporal convolutions along time. The paper re-formulates the original whole-sequence DGNN into an online, per-frame classifier: given a 0.5 s window of skeleton data, it predicts the intention label for a 0.25 s offset, and a 15-sample averaging filter smooths the output. The assistive controller then sets the desired robot force to $f_d = i_h f_{com}$, where $i_h \in \{-1,0,+1\}$ is the predicted intention and $f_{com}$ is a friction-compensation force found in a one-time object-exploration step using the robot's force-torque sensor.

What would settle it

Repeat the six experimental conditions with $f_{com}$ deliberately detuned, for instance at 50% of the manually tuned value: if the mean and cumulative human force in the assisted trials no longer drop relative to the dry trials, the central claim fails and the reported reduction is an artifact of calibration rather than of intention prediction.

Watch

Extended reading notes

Core claim

The central claim is that collaborative pushing and pulling of heavy objects on frictional surfaces can be assisted without any direct measurement of human force: the human's push/pull/idle intention is predicted from a time window of skeleton joint and bone data, and the robot applies force in the predicted direction at a magnitude that compensates for static friction. The authors report 93.4% accuracy and 94.1% balanced accuracy on the intention classifier, and show that assisted trials reduce the human's mean and cumulative force relative to dry trials in most of the six experimental conditions. The key sensor-side argument is that the robot's force-torque sensor measures $f_r = -f_h + f_f$, so human force cannot be decoupled from friction; posture-based intention prediction sidesteps this and lets assistance begin before static friction is overcome.

Load-bearing premise

The effort reduction depends on the manually tuned compensation force $f_{com}$, found by adjusting the robot force until it could almost move the object by itself; if that value is mistuned, the same experiments could show much less or no benefit.

Editorial extensions

If this is right

  • A robot can assist physical collaborative manipulation using only skeleton tracking and its own wrist force sensor, with no sensor mounted on the object.
  • Because intention is predicted before significant human force appears, assistance can ramp up before static friction is broken, which force-only controllers cannot do.
  • The reported benefit increases with object mass: for the heavier 36.0 kg box the assisted effort stays flat while the dry effort rises, so the method is most valuable where human fatigue is highest.
  • Decreasing motion speed slightly increases assisted effort because of prediction delay and a low force-control gain, indicating the controller, not the intention predictor, sets the speed limit.

Reading between the lines

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

  • The manually tuned $f_{com}$ is the least controlled element of the system; replacing it with an online friction estimator would presumably make the approach robust to surface changes, but the paper does not test this.
  • The 0.25 s prediction horizon plus 75 ms filter delay means the robot's force often leads the human's force; a force-only baseline cannot reproduce this lead time because static friction hides the human force until motion starts.
  • With only two participants, the classifier may have learned person-specific posture cues; a cross-subject evaluation would tell whether the learned push/pull kinematics generalize.
  • The matrix-form formulation suggests the same controller could handle planar pushing if the y and z intention dimensions were populated; predicting planar intent from skeleton data is the natural next test.
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

4 major / 7 minor

Summary. The manuscript presents a context-aware assistance system for a human and a robot collaboratively pushing/pulling heavy objects on a frictional surface. Human intention (push/pull/idle) is predicted online from Xsens skeleton data with a Directed Graph Neural Network (DGNN), and the robot applies a desired force f_d = i_h f_com, where i_h is the predicted intention and f_com is an object-specific compensation force determined by a one-time robot-only exploration. The system is evaluated in six experiments with two participants, comparing an assisted condition against a 'dry' condition in which the robot is excluded. The authors report reduced mean and cumulative human handle force in most conditions and conclude that posture-based context recognition improves robot decision-making and control efficiency. The paper is positioned as a proof of concept for indirect pHRI where direct force feedback on the object is unavailable.

Significance. If the claims are established, the work addresses a practically relevant gap: collaborative sliding of heavy objects where human intention cannot be read from an object-mounted force sensor. The paper contributes a novel task formulation, a real-robot experimental protocol, and a public video. A strength is that the DGNN is trained on separate robot-free data and then used online, so the main positive result is not circular with respect to training. However, the central scientific claim—that the skeleton-based intention predictor, rather than the large manually tuned assistive force, is responsible for the effort reduction—is not isolated by the current experiments. No alternative intention source, no ablation of the DGNN, and no sensitivity analysis of f_com are provided. The statistical comparison is also potentially biased by the 15 N force threshold used to discard frames. The significance is therefore conditional: the paper is a promising proof of concept, but its abstract-level claims about posture-based context recognition enhancing decision-making require additional control experiments.

major comments (4)
  1. [Sec. III-C, Sec. IV-B, Fig. 8] See above.
  2. [Sec. IV-B, Eq. (2), Table I] See above.
  3. [Sec. IV-B, Fig. 7, Fig. 8] See above.
  4. [Sec. IV-B, Sec. IV-C] See above.
minor comments (7)
  1. [Sec. III-B]
  2. [Sec. IV-B]
  3. [Sec. IV-A]
  4. [Fig. 8]
  5. [Sec. IV]
  6. [Sec. IV-C]
  7. [Sec. IV]

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the main claimed result is an empirical, externally benchmarked validation, and the self-citations are not load-bearing.

full rationale

The paper is an empirical validation rather than a derivation. The skeleton-based intention predictor is trained on a separate robot-free dataset and validated to 93.4% accuracy / 94.1% balanced accuracy against labels derived from object velocity, which is an external benchmark not constructed from the robot's assistance law. The assistance law f_d = i_h * f_com (Sec. III-D, Eq. 2) uses a manually tuned f_com, but the paper does not present the human-effort reduction as a prediction from f_com; it is a measured comparison against a no-robot dry baseline. The absence of an ablation isolating the skeleton-based predictor is a real limitation, and the paper itself concedes this in Sec. V ('each component of the system can be further validated'), but that is a question of component-level attribution, not a reduction of the claimed result to its own inputs. Self-citations such as [20], [26], [28], and [29] appear in related work and are not load-bearing for the central claim. No uniqueness theorem is imported from the authors' prior work, and no ansatz is smuggled in via self-citation. Therefore no circular step is exhibited under the stated criteria.

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

The central result depends on several hand-tuned quantities, most importantly the per-object compensation force fcom and the evaluation's 15 N filter threshold. The quasi-static force model and the discrete 3-class intention representation are domain assumptions that limit the scope. No new physical entities are introduced.

free parameters (5)
  • fcom (compensation force) = 65 N for 27.7 kg box, 80 N for 36.0 kg box
    Determined interactively in an 'object exploration' step by iterating the robot force until it 'could almost move the object by itself'. Used as the desired force magnitude f_d = ih * fcom, so the level of assistance is directly set by this tuned value.
  • Force filter threshold = 15 N
    Frames with human-side force below 15 N are removed from the effort calculation to filter out idle time. This threshold is arbitrary and can affect the comparison between dry and assisted conditions.
  • Intention prediction window size = 0.5 s
    Chosen as the input time window for the DGNN; affects what postural cues are available for prediction.
  • Prediction offset = 0.25 s
    The time between the end of the input window and the target label; directly related to the claimed anticipation of 1.54 s for participant 1.
  • DGNN hyperparameters = 3 GTN layers with 32, 64, 64 channels; dropout 0.3; learning rate 0.02; weight decay 0.005
    Manually selected based on validation balanced accuracy; a different architecture could change prediction performance.
assumptions (5)
  • domain assumption Quasi-static force balance f_r = -f_h + f_f (Eq. 1)
    Assumes no inertial effects and Coulomb friction so the robot-side force is the net of human force and friction; used to justify the infeasibility of admittance control. May not hold during dynamic motion or with velocity-dependent friction.
  • domain assumption Human intention is represented by three discrete labels (push, pull, idle) along one axis
    The controller directly maps the predicted class to a force direction; this discretization limits the task to 1-DoF and ignores gradual or multi-directional intentions.
  • domain assumption Skeleton pose from Xsens reflects intention before force onset
    The whole method relies on posture carrying anticipatory information; the reported lead time is large for participant 1 but near zero for participant 2, so the premise is not uniformly supported.
  • domain assumption Training data collected without the robot is representative of the human's pose during robot-assisted interaction
    The DGNN is trained on dry pushes and pulls, but deployed while the robot applies force, which may change the human's posture and effort. Online accuracy is not reported.
  • domain assumption The statically-determined friction compensation force fcom remains appropriate during dynamic motion
    The controller applies the constant fcom in both static and moving phases; the paper acknowledges this can lead to more help than friction compensation in dynamic cases, but the reported benefit depends on this value.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Context-aware collaborative pushing of heavy objects using skeleton-based intention prediction." pith.science (2026). https://pith.science/paper/SMM27SPV

@misc{pith2026250510239,
  author       = {Pith},
  title        = {Pith review of: Context-aware collaborative pushing of heavy objects using skeleton-based intention prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SMM27SPV}},
  note         = {Machine review of arXiv:2505.10239}
}
read the original abstract

In physical human-robot interaction, force feedback has been the most common sensing modality to convey the human intention to the robot. It is widely used in admittance control to allow the human to direct the robot. However, it cannot be used in scenarios where direct force feedback is not available since manipulated objects are not always equipped with a force sensor. In this work, we study one such scenario: the collaborative pushing and pulling of heavy objects on frictional surfaces, a prevalent task in industrial settings. When humans do it, they communicate through verbal and non-verbal cues, where body poses, and movements often convey more than words. We propose a novel context-aware approach using Directed Graph Neural Networks to analyze spatio-temporal human posture data to predict human motion intention for non-verbal collaborative physical manipulation. Our experiments demonstrate that robot assistance significantly reduces human effort and improves task efficiency. The results indicate that incorporating posture-based context recognition, either together with or as an alternative to force sensing, enhances robot decision-making and control efficiency.

Figures

Figures reproduced from arXiv: 2505.10239 by the authors.

Figure 1
Figure 1. The task is split into three possible movements: when the human intends to push, pull, or stay idle. The collaborative robot predicts the intended motion from the human skeleton tracking data to act appropriately in the context. in [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The information flow in our system. The independent bodies of the human, object, and robot interact by direct forces f h , f o and fr . Human intention is predicted using human posture sensing, and object friction information is discovered using the robot-side F/T sensor. Intention vector ih and compensation force fcom are used by the assistance controller to generate the desired robot forces f d that are aligned wi… view at source ↗
Figure 3
Figure 3. Given a time window of input data X, compute the human intention label Y after a given time offset. This can be modeled as a supervised learning problem (classification or regression). In this work, we use the skeleton data Xh as the input because it preserves the meaningful structure of the human posture information. Xh includes the joint (3-D position) and the bone data (difference between the connected joint posi… view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: a) The user wears the Xsens markers for skeleton tracking; b) An F/T sensor is attached between the human-side handle and the box; c) An Optitrack marker is attached to the object for tracking its pose xb; d) A rigid wooden box filled with heavy items; e) The robot gra…
Figure 6
Figure 6. Figure 6: Comparison of the force measurements along the push direction with reference to the intention prediction during exp 2. Cyan areas indicate the real-time pull and push prediction. The time series are shifted to align horizontally. the box velocity ˙xb, and used it to de…
Figure 7
Figure 7. Figure 7: We also show the mean value of the force during that experiment. Then, we separately calculate the average and cumulative force for each pull/push action. Mean force and the cumulative force (area under the curve) values are given for all experiments in [PITH_FULL_IMA…
Figure 8
Figure 8. Figure 8: Mean and cumulative force exerted by the human in 10 repeated pull or push actions of each experiment case. Please see Table I for the experiment cases. ∗ indicates statistical significance according to two-sample t-test with unequal variances (p < 0.01). ues below 15 …
Figure 9
Figure 9. Figure 9: Robot-side force fr measurements from exp 4 during two pull and two push motions. The force direction is inverse of the human-side force that is shown in [PITH_FULL_IMAGE:figures/full_fig_p006_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages

  1. [1]

    Analyzing human–human interactions: A survey,

    A. Stergiou and R. Poppe, “Analyzing human–human interactions: A survey,” Computer Vision and Image Understanding , vol. 188, p. 102799, 11 2019

    work page 2019

  2. [2]

    Collision-free human-robot collaboration based on context awareness,

    H. Liu and L. Wang, “Collision-free human-robot collaboration based on context awareness,” Robotics and Computer-Integrated Manufac- turing, vol. 67, p. 101997, 2 2021

    work page 2021

  3. [3]

    Progress and prospects of the human-robot collabo- ration,

    A. Ajoudani, A. M. Zanchettin, S. Ivaldi, A. Albu-Sch ¨affer, K. Kosuge, and O. Khatib, “Progress and prospects of the human-robot collabo- ration,” Autonomous Robots , vol. 42, no. 5, pp. 957–975, 2018

    work page 2018

  4. [4]

    Human body mechanics of pushing and pulling: Analyzing the factors of task- related strain on the musculoskeletal system,

    A. Argubi-Wollesen, B. Wollesen, M. Leitner, and K. Mattes, “Human body mechanics of pushing and pulling: Analyzing the factors of task- related strain on the musculoskeletal system,” Safety and Health at Work, vol. 8, pp. 11–18, 3 2017

    work page 2017

  5. [5]

    Ergonomic evaluation of workers during manual material handling,

    M. Rajendran, A. Sajeev, R. Shanmugavel, and T. Rajpradeesh, “Ergonomic evaluation of workers during manual material handling,” Materials Today: Proceedings , vol. 46, pp. 7770–7776, 2021

    work page 2021

  6. [6]

    Human-humanoid collaborative carrying,

    D. J. Agravante, A. Cherubini, A. Sherikov, P.-B. Wieber, and A. Kheddar, “Human-humanoid collaborative carrying,” IEEE Trans- actions on Robotics , vol. 35, pp. 833–846, 8 2019

    work page 2019

  7. [7]

    Choosing poses for force and stiffness control,

    A. Ajoudani, N. G. Tsagarakis, and A. Bicchi, “Choosing poses for force and stiffness control,” IEEE Transactions on Robotics , vol. 33, pp. 1483–1490, 12 2017

    work page 2017

  8. [8]

    Skeleton-based action recognition with directed graph neural networks,

    L. Shi, Y . Zhang, J. Cheng, and H. Lu, “Skeleton-based action recognition with directed graph neural networks,” Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2019-June, pp. 7904–7913, 2019

    work page 2019

Show all 34 references
  1. [9]

    A review of intent detection, arbitration, and communication aspects of shared control for physical human–robot interaction,

    D. P. Losey, C. G. McDonald, E. Battaglia, and M. K. O’Malley, “A review of intent detection, arbitration, and communication aspects of shared control for physical human–robot interaction,” Applied Mechanics Reviews, vol. 70, 1 2018

    2018

  2. [10]

    Context-aware hand gesture interaction for human–robot collaboration in construction,

    X. Wang, D. Veeramani, F. Dai, and Z. Zhu, “Context-aware hand gesture interaction for human–robot collaboration in construction,” Computer-Aided Civil and Infrastructure Engineering , 4 2024

    2024

  3. [11]

    A cyber-physical context- aware system for coordinating human-robot collaboration,

    N. Nikolakis, K. Sipsas, and S. Makris, “A cyber-physical context- aware system for coordinating human-robot collaboration,” Procedia CIRP, vol. 72, pp. 27–32, 2018

    2018

  4. [12]

    Toward a context-aware human–robot interaction framework based on cognitive development,

    J. Quintas, G. S. Martins, L. Santos, P. Menezes, and J. Dias, “Toward a context-aware human–robot interaction framework based on cognitive development,” IEEE Transactions on Systems, Man, and Cybernetics: Systems , vol. 49, pp. 227–237, 1 2019

    2019

  5. [13]

    Learning and blending robot hugging behaviors in time and space,

    M. Drolet, J. Campbell, and H. B. Amor, “Learning and blending robot hugging behaviors in time and space,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2023, pp. 12 071–12 077

    2023

  6. [14]

    Vision-and tactile-based continuous multimodal intention and attention recognition for safer physical human–robot interaction,

    C. Y . Wong, L. Vergez, and W. Suleiman, “Vision-and tactile-based continuous multimodal intention and attention recognition for safer physical human–robot interaction,” IEEE Transactions on Automation Science and Engineering , vol. 21, pp. 3205–3215, 2023

    2023

  7. [15]

    Expanding humanoid’s material-handling capabilities using capture point walking,

    J. C. Vaz and P. Y . Oh, “Expanding humanoid’s material-handling capabilities using capture point walking,” Proceedings of the American Control Conference, vol. 2020-July, pp. 2082–2087, 7 2020

    2020

  8. [16]

    Multi-contact heavy object pushing with a centaur-type humanoid robot: Planning and control for a real demonstrator,

    M. P. Polverini, A. Laurenzi, E. M. Hoffman, F. Ruscelli, and N. G. Tsagarakis, “Multi-contact heavy object pushing with a centaur-type humanoid robot: Planning and control for a real demonstrator,” IEEE Robotics and Automation Letters , vol. 5, pp. 859–866, 4 2020

    2020

  9. [17]

    Whole-body pushing manipulation with contact posture planning of large and heavy object for humanoid robot,

    M. Murooka, S. Nozawa, Y . Kakiuchi, K. Okada, and M. Inaba, “Whole-body pushing manipulation with contact posture planning of large and heavy object for humanoid robot,” Proceedings - IEEE International Conference on Robotics and Automation , vol. 2015-June, pp. 5682–5689, 5 2015

    2015

  10. [18]

    Hierarchical deep reinforce- ment learning to drag heavy objects by adult-sized humanoid robot,

    S. Saeedvand, H. Mandala, and J. Baltes, “Hierarchical deep reinforce- ment learning to drag heavy objects by adult-sized humanoid robot,” Applied Soft Computing , vol. 110, p. 107601, 10 2021

    2021

  11. [19]

    Dynamic pushing strategies for dynamically stable mobile manipulators,

    P. Kolhe, N. Dantam, and M. Stilman, “Dynamic pushing strategies for dynamically stable mobile manipulators,” Proceedings - IEEE International Conference on Robotics and Automation , pp. 3745–3750, 2010

    2010

  12. [20]

    Pushing in the dark: A reactive pushing strategy for mobile robots using tactile feedback,

    I. Ozdamar, D. Sirintuna, R. Arbaud, and A. Ajoudani, “Pushing in the dark: A reactive pushing strategy for mobile robots using tactile feedback,” IEEE Robotics and Automation Letters , vol. 9, no. 8, pp. 6824–6831, 2024

    2024

  13. [21]

    Linear time-varying mpc for nonprehensile object manipulation with a nonholonomic mobile robot,

    F. Bertoncelli, F. Ruggiero, and L. Sabattini, “Linear time-varying mpc for nonprehensile object manipulation with a nonholonomic mobile robot,” Proceedings - IEEE International Conference on Robotics and Automation, pp. 11 032–11 038, 2020

    2020

  14. [22]

    A con- vex polynomial force-motion model for planar sliding: Identification and application,

    J. Zhou, R. Paolini, J. A. Bagnell, and M. T. Mason, “A con- vex polynomial force-motion model for planar sliding: Identification and application,” Proceedings - IEEE International Conference on Robotics and Automation , vol. 2016-June, pp. 372–377, 2016

    2016

  15. [23]

    Let’s push things forward: A survey on robot pushing,

    J. St ¨uber, C. Zito, and R. Stolkin, “Let’s push things forward: A survey on robot pushing,” Frontiers in Robotics and AI , vol. 7, 2 2020

    2020

  16. [24]

    Contact-implicit planning and control for non-prehensile manipulation using state- triggered constraints,

    M. Wang, A. ¨O. ¨Onol, P. Long, and T. Padır, “Contact-implicit planning and control for non-prehensile manipulation using state- triggered constraints,” in Robotics Research , A. Billard, T. Asfour, and O. Khatib, Eds. Cham: Springer Nature Switzerland, 2023, pp. 189–204

    2023

  17. [25]

    Diffusion co-policy for synergistic human-robot collaborative tasks,

    E. Ng, Z. Liu, and M. Kennedy, “Diffusion co-policy for synergistic human-robot collaborative tasks,” IEEE Robotics and Automation Letters, vol. 9, pp. 215–222, 1 2024

    2024

  18. [26]

    Enhancing human–robot collaborative transportation through obstacle-aware vibrotactile warning and virtual fixtures,

    D. Sirintuna, T. Kastritsi, I. Ozdamar, J. M. Gandarias, and A. Ajoudani, “Enhancing human–robot collaborative transportation through obstacle-aware vibrotactile warning and virtual fixtures,” Robotics and Autonomous Systems , vol. 178, p. 104725, 2024. [Online]. Available: ht...

    2024

  19. [27]

    Recognition of haptic interaction patterns in dyadic joint object manipulation,

    C. E. Madan, A. Kucukyilmaz, T. M. Sezgin, and C. Basdogan, “Recognition of haptic interaction patterns in dyadic joint object manipulation,” IEEE Transactions on Haptics , vol. 8, pp. 54–66, 1 2015

    2015

  20. [28]

    Carrying the uncarri- able: a deformation-agnostic and human-cooperative framework for unwieldy objects using multiple robots,

    D. Sirintuna, I. Ozdamar, and A. Ajoudani, “Carrying the uncarri- able: a deformation-agnostic and human-cooperative framework for unwieldy objects using multiple robots,” 2023 IEEE International Conference on Robotics and Automation (ICRA) , pp. 7497–7503, 5 2023

    2023

  21. [29]

    An object deformation-agnostic framework for human–robot collaborative trans- portation,

    D. Sirintuna, A. Giammarino, and A. Ajoudani, “An object deformation-agnostic framework for human–robot collaborative trans- portation,” IEEE Transactions on Automation Science and Engineer- ing, vol. 21, pp. 1986–1999, 4 2024

    1986

  22. [30]

    Friction compensation in robotics: an overview,

    B. Bona and M. Indri, “Friction compensation in robotics: an overview,” Proceedings of the 44th IEEE Conference on Decision and Control, pp. 4360–4367, 2005

    2005

  23. [31]

    Human action recognition from various data modalities: A review,

    Z. Sun, Q. Ke, H. Rahmani, M. Bennamoun, G. Wang, and J. Liu, “Human action recognition from various data modalities: A review,” IEEE transactions on pattern analysis and machine intelligence , vol. 45, no. 3, pp. 3200–3225, 2022

    2022

  24. [32]

    A smoothed gms friction model suited for gradient-based friction state and parameter estimation,

    M. Boegli, T. D. Laet, J. D. Schutter, and J. Swevers, “A smoothed gms friction model suited for gradient-based friction state and parameter estimation,” IEEE/ASME Transactions on Mechatronics, vol. 19, pp. 1593–1602, 10 2014. [Online]. Available: http://ieeexplore.ieee.org/do...

    2014

  25. [33]

    Observer of nonlinear friction dynam- ics for motion control,

    M. Ruderman and M. Iwasaki, “Observer of nonlinear friction dynam- ics for motion control,” IEEE Transactions on Industrial Electronics , vol. 62, pp. 5941–5949, 9 2015

    2015

  26. [34]

    The balanced accuracy and its posterior distribution,

    K. H. Brodersen, C. S. Ong, K. E. Stephan, and J. M. Buhmann, “The balanced accuracy and its posterior distribution,” in 2010 20th international conference on pattern recognition . IEEE, 2010, pp. 3121–3124

    2010

Pith tools

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