Pith. sign in

REVIEW 3 major objections 5 minor 40 references

Hearing the Slide: Acoustic-Guided Constraint Learning for Fast Non-Prehensile Transport

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A learned, velocity-dependent friction constraint, inferred from contact-microphone detections of slide onset, reduces object displacement in fast tray transport by an average of 86.0% compared with the standard Coulomb friction baseline.

desk verdict Promising acoustic friction-learning pipeline, but the experiments don't yet rule out that a constant conservative friction scaling would do just as well. read the letter →

arxiv 2506.09169 v1 pith:UXQLXPFH submitted 2025-06-10 cs.RO

Yuemin Mao , Bardienus P. Duisterhof , Moonyoung Lee , Jeffrey Ichnowski This is my paper · ORCID
classification cs.RO
keywords non-prehensiletransportacousticsensingcontactmicrophonefrictionmodellearningtime-optimalmotionplanningCoulombobjecttrajectoryoptimization
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

This paper tries to establish that a robot carrying objects on a tray can move them fast without letting them slide if the friction limit used by motion planning is learned from acoustic signals rather than fixed by the Coulomb model. The method attaches a contact microphone to the tray, detects the moment an object starts to slide during training motions, and converts those onsets into a velocity-dependent friction scale $\alpha = g(\|\mathbf{v}\|)$ inside the planner's safety constraint. On a UR5e with eight objects and twelve configurations, the learned constraint reduces average object displacement by 53.9% with one training model and 86.0% with another, relative to the standard Coulomb baseline, and it also outperforms an equal-duration Coulomb ablation. The practical point is that cheap audio feedback can capture speed-dependent vibration effects that a static friction coefficient misses.

What carries the argument

The machinery is the learned dynamic friction parameter $\alpha = g(\|\mathbf{v}\|)$: a scalar multiplier on the measured static friction coefficient $\mu_s$, trained from acoustic slide-onset detections and evaluated inside the trajectory optimizer's non-sliding constraint at each time step. The constraint takes the form $\|\alpha_i \mu_s (\mathbf{a}_i \cdot \mathbf{n}_i)\| \ge \|\mathbf{a}_i - (\mathbf{a}_i \cdot \mathbf{n}_i)\mathbf{n}_i\|$, where $\mathbf{a}_i$ is inertial acceleration at the object centroid and $\mathbf{n}_i$ is the tray normal. The learned $\alpha$ decreases with velocity, tightening the friction limit when vibration is strongest; this is what lets the planner trade speed against slide risk.

What would settle it

Run a held-out object on an arced trajectory that reaches speeds near the top of the training range while filming the tray with a high-speed camera; if the learned $\alpha = g(\|\mathbf{v}\|)$ allows accelerations that make the object slide more than the Coulomb baseline before it falls, the central claim is falsified.

Watch

Extended reading notes

Core claim

The author's claim, stated on their own terms: the standard Coulomb constraint $F_f \le \mu_s F_n$ is imprecise at high speed because mechanical vibration grows with tray velocity and reduces the effective friction, so objects slide even when the planner believes they are safe. The paper replaces the constant $\mu_s$ with $\tilde{\mu} = \alpha \mu_s$, where $\alpha = g(\|\mathbf{v}\|)$ is a learned function of the tray's linear speed. Slide onset is detected from filtered contact-microphone spectrograms of straight-line horizontal trials, and the resulting pairs $(\|\mathbf{v}_{\text{sliding}}\|, \|\mathbf{a}_{\text{sliding}}\|)$ train an MLP that makes $\alpha$ smaller as velocity increases. The same constraint is then enforced at every discretized step of an optimization-based time-optimal planner, and experiments show that this learned constraint keeps objects on the tray during fast arced motions, cutting displacement by up to 86.0% compared with the Coulomb baseline while beating an equal-duration Coulomb ablation.

Load-bearing premise

The load-bearing premise is that the sliding boundary learned from straight-line horizontal motions of two training objects, encoded as $\alpha = g(\|\mathbf{v}\|)$, transfers to arced trajectories and to the other eight objects, contact surfaces, and multi-object configurations without retraining.

Editorial extensions

If this is right

  • If the result holds, any tray-carrying manipulator can use one contact microphone as a low-cost high-frequency sensor to build a motion-specific friction constraint without cameras or tactile arrays.
  • The learned constraint beats an equal-duration Coulomb ablation, so the displacement improvements are not simply a consequence of moving slower.
  • Practitioners can explicitly trade transport time against residual displacement by choosing more or less conservative learned friction models.
  • The method extends to multi-object configurations and irregularly shaped objects, provided contact surfaces are even and objects do not topple or interact.
  • The learned parameter is tied to a specific robot's vibration signature, so the same trained function would need recalibration when transferred to a different manipulator.

Reading between the lines

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

  • The velocity-only parameterization likely misses acceleration-direction effects such as centripetal and Coriolis forces on arcs; a richer feature space including jerk or normal acceleration could improve accuracy on sharp curves while keeping the same acoustic setup.
  • The same slide-onset detection could serve online slip monitoring: instead of only planning with a learned $\alpha$, the microphone could trigger replanning if sliding begins mid-trajectory.
  • Because $\alpha$ shrinks with speed, it implicitly encodes vibration amplitude; directly measuring tray vibration RMS from the same microphone might let $\alpha$ generalize across objects without retraining.
  • A similar procedure could learn constraints for other contact-rich tasks such as pushing, scooping, or in-hand manipulation, where friction transitions are audible.
Share X Bluesky LinkedIn Reddit HN

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 learning a velocity-dependent friction scaling factor alpha = g(||v||) from contact-microphone detections of the onset of object sliding during straight-line horizontal tray motions, and embedding it into the GOMP-FIT optimization-based planner as a non-sliding constraint. The learned factor is intended to capture vibration-induced reductions in effective friction that the standard Coulomb model misses during fast transport. Physical experiments on a UR5e compare trajectories generated with the standard Coulomb model, two learned models (trained on PLA and glass objects), and two equal-duration ablations, measuring object displacement after transport for 12 object configurations. The authors report that the more conservative learned model reduces mean displacement by 86.0% relative to the Coulomb baseline and that the learned models outperform the equal-duration Coulomb ablation in most cases.

Significance. If the central claim holds, the contribution is practically significant: it offers a low-cost sensing modality and a straightforward way to make fast non-prehensile transport safer. The paper's strengths include physical experiments with direct displacement measurement, a sensible equal-duration ablation intended to separate speed reduction from constraint quality, a diverse object set with explicit notes on assumption violations, and careful enumeration of modeling assumptions. The main weakness is that the equal-duration ablation does not control for a constant friction multiplier, so the velocity-dependence of the learned model is not actually isolated. Given that the learned alpha is at most 1 and decreases with speed, the improved performance could be explained by a more conservative constant bound. This, together with the absence of variance reporting and the training/evaluation domain gap, means the evidence currently supports 'a more conservative learned constraint reduces sliding' more strongly than 'acoustic, velocity-dependent friction conditioning is necessary'.

major comments (3)
  1. [§VI.B/Table I] The AS-Coulomb ablation controls for trajectory duration but not for the magnitude of the friction multiplier. Because the learned alpha is identically 1 at ||v||=0 and decreases with velocity (Fig. 5), Learned-Glass could outperform AS-Coulomb simply by enforcing a more conservative constant bound, with no velocity dependence. The sentence in Section VI.C claiming that the ablation rules out 'scaling the measured friction coefficient by a constant α<1' is therefore not supported by the reported experiments. Please add an ablation with a constant α (e.g., the mean or a selected quantile of the learned α) at the same duration as Learned-Glass, or otherwise vary α independently of velocity, to isolate the effect of velocity conditioning.
  2. [§VI.B/Table I] No trial-level variance or significance testing is reported. Each entry is a mean over 5 trials (7 for EGAD), but standard deviations, per-trial displacements, and paired comparisons are absent. Without these, the headline 86.0% reduction and the smaller differences between Learned-Glass and AS-Coulomb (e.g., 1.7 vs 2.3 mm for glass big) cannot be distinguished from experimental noise. Please report per-trial data and appropriate statistics.
  3. [§V.C/§VI.B] The learned model is trained on straight-line, level-tray, single-object horizontal motions with a bounded tray, but evaluated on arced trajectories with Coriolis/centrifugal forces, a boundary-free tray with protective film, eight objects, and multi-object configurations. Since alpha is a function of ||v|| only, the transfer to these conditions is an assumption. Please provide evidence that the learned alpha predicts slide onset in the evaluation setting (e.g., compare predicted and observed slide-onset velocities/accelerations) or explicitly bound the generalization claim.
minor comments (5)
  1. [§V.C] The symbol g is used both for gravity and for the learned function g(||v||) in the same derivation; please use separate symbols to avoid confusion.
  2. [§V.C] The criterion for 'significantly greater magnitude' when detecting t_sliding is not defined; a reproducible threshold or detection algorithm should be specified.
  3. [Table II] The row label 'Learned-Glass, AS' is ambiguous; clarify whether AS-Coulomb and AS-No Constraints share the same duration as Learned-Glass and indicate this directly in the table.
  4. [§VI.C] The description of the two outliers is unclear: for the sponge soft side, AS-Coulomb (1.1 mm) is actually better than Learned-Glass (1.7 mm), so the statement that AS-Coulomb performs worse in all non-outlier cases should be reconciled with Table I.
  5. [Fig. 5] The learned alpha curves have no uncertainty bands despite being fit to only 85 and 70 data points; adding validation error or confidence intervals would help assess overfitting.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the learned friction coefficient is a calibrated threshold from slide-onset data, and the displacement outcome is independently measured.

full rationale

The paper's central claim is that a velocity-dependent friction multiplier, learned from acoustic slide-onset detections, reduces measured object displacement. This is not circular: the training data provide slide-onset acceleration thresholds, which are converted into alpha values via alpha_training = ||a_sliding||/(mu_s ||g||), and those alpha values then appear as a constraint multiplier in Eq. (2). The learned alpha is a calibrated empirical threshold, not a quantity defined in terms of the final displacement metric. The displacement reductions in Table I are measured from real robot runs after planning, so no equation reduces the reported outcome to a fitted input by construction. The self-citations to GOMP-FIT and GOMP-ST are methodological references to a motion-planning solver and a data-collection pipeline; they do not smuggle in the paper's conclusion or forbid alternative friction models. The equal-duration ablation does not include a constant-alpha<1 condition, so the claim that the learned model captures more than a constant scaling is not fully controlled; however, this is an experimental-design limitation, not a circularity, because no fitted displacement value is fed back into the learned constraint. Overall, the derivation chain is self-contained and the key evaluation is an independent physical measurement.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

All claims rest on measured static friction coefficients, a learned velocity-only scaling function, and the assumption that the same speed-dependent alpha applies across objects and trajectory shapes. The paper's own assumptions in Section III are explicit, but several are violated by the test set, including the sponge, multi-object configurations, and thin objects, and the training and evaluation tray setups differ.

free parameters (2)
  • alpha = g(||v||) (learned friction scaling function) = MLP weights trained on 85 (PLA) and 70 (glass) slide-onset pairs
    The central constraint Eq. 2 uses alpha_i = g(||v_i||) in place of mu_s; the function is fitted to data from two training objects and is the main learned quantity.
  • Synthetic data point at ||v||=0 = alpha = 1 (a_sliding = mu_s g)
    Section V.C: data are augmented by assuming a_sliding = mu_s g when velocity is zero and linearly interpolating to the minimum identified v_sliding; this assumed endpoint shapes the learned curve.
assumptions (5)
  • domain assumption Objects have even contact surfaces, do not topple, experience all external forces at their centroids, and do not interact with each other.
    Stated in Section III Problem Statement; violated deliberately by the sponge (uneven contact) and multi-object trials, yet the learned constraint is applied to all.
  • domain assumption During straight-line horizontal data collection, only horizontal inertial force and gravity act on the object at slide onset, giving alpha_training = ||a_sliding||/(mu_s ||g||).
    Section V.C; ignores vertical acceleration, Coriolis/centrifugal, and vibration-induced force fluctuations during the training trials.
  • ad hoc to paper The effective friction coefficient is the same everywhere on the tray at a given time and depends on the robot state only through the magnitude of linear velocity.
    Section V.B states tilde-mu is constant anywhere on the tray and Section V.C trains a function alpha=g(||v||); this collapses object, material, acceleration, and normal-force dependence into one speed-dependent scalar.
  • domain assumption Acoustic signal from an empty tray can be treated as non-stationary noise to be subtracted from signals with the object, and the first significantly greater spectral magnitude marks slide onset.
    Section V.C data-processing pipeline; the threshold for significant is not quantitatively defined, so the training labels rely on this subjective segmentation.
  • standard math Trajectory optimization with the SQP solver from GOMP-FIT yields a locally time-optimal trajectory that satisfies the discretized constraints.
    Section IV.A; the planner is taken as a tool, but local optimality and discrete-time constraint satisfaction are assumed to represent continuous-time safety.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hearing the Slide: Acoustic-Guided Constraint Learning for Fast Non-Prehensile Transport." pith.science (2026). https://pith.science/paper/UXQLXPFH

@misc{pith2026250609169,
  author       = {Pith},
  title        = {Pith review of: Hearing the Slide: Acoustic-Guided Constraint Learning for Fast Non-Prehensile Transport},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UXQLXPFH}},
  note         = {Machine review of arXiv:2506.09169}
}
read the original abstract

Object transport tasks are fundamental in robotic automation, emphasizing the importance of efficient and secure methods for moving objects. Non-prehensile transport can significantly improve transport efficiency, as it enables handling multiple objects simultaneously and accommodating objects unsuitable for parallel-jaw or suction grasps. Existing approaches incorporate constraints based on the Coulomb friction model, which is imprecise during fast motions where inherent mechanical vibrations occur. Imprecise constraints can cause transported objects to slide or even fall off the tray. To address this limitation, we propose a novel method to learn a friction model using acoustic sensing that maps a tray's motion profile to a dynamically conditioned friction coefficient. This learned model enables an optimization-based motion planner to adjust the friction constraint at each control step according to the planned motion at that step. In experiments, we generate time-optimized trajectories for a UR5e robot to transport various objects with constraints using both the standard Coulomb friction model and the learned friction model. Results suggest that the learned friction model reduces object displacement by up to 86.0% compared to the baseline, highlighting the effectiveness of acoustic sensing in learning real-world friction constraints.

Figures

Figures reproduced from arXiv: 2506.09169 by the authors.

Figure 1
Figure 1. Fast non-prehensile transport. Top left: Time-optimized transport without constraints drops the object. Bottom left: With friction-based constraints, the motion planner initiates tray tilt to compensate for inertia effects, safely transporting multiple fragile objects. Right: We propose a learned dynamic constraint to balance the inertial and frictional forces during fast transport, reducing object sliding. limit or… view at source ↗
Figure 2
Figure 2. Raw contact-microphone signal collected when a UR5e robot’s tray [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. (a) Data collection setup. (b) Contact microphone attached under the [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Visualization of the models trained for dynamic constraint parameter [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 33 canonical work pages

  1. [1]

    Dex-Net 1.0: A cloud-based network of 3D objects for robust grasp planning using a multi-armed bandit model with correlated rewards

    J. Mahler, F. T. Pokorny, B. Hou, M. Roderick, M. Laskey, M. Aubry, K. Kohlhoff, . T. Kr ¨oger, J. Kuffner, and K. Goldberg, “Dex-Net 1.0: A cloud-based network of 3D objects for robust grasp planning using a multi-armed bandit model with correlated rewards.” IEEE, 2016, pp. 1957–1964

    work page 2016

  2. [2]

    Adversarial grasp objects,

    D. Wang, D. Tseng, P. Li, Y . Jiang, M. Guo, M. Danielczuk, J. Mahler, J. Ichnowski, and K. Goldberg, “Adversarial grasp objects,” in2019 IEEE 15th International Conference on Automation Science and Engineering (CASE), 2019, pp. 241–248

    work page 2019

  3. [3]

    Egad! an evolved grasping analysis dataset for diversity and reproducibility in robotic manipulation,

    D. Morrison, P. Corke, and J. Leitner, “Egad! an evolved grasping analysis dataset for diversity and reproducibility in robotic manipulation,” 2020. [Online]. Available: https://arxiv.org/abs/2003. 01314

    work page 2020

  4. [4]

    Dex- Net 3.0: Computing robust vacuum suction grasp targets in point clouds using a new analytic model and deep learning

    J. Mahler, M. Matl, X. Liu, A. Li, D. Gealy, and K. Goldberg, “Dex- Net 3.0: Computing robust vacuum suction grasp targets in point clouds using a new analytic model and deep learning.” IEEE, 2018, pp. 1–8

    work page 2018

  5. [5]

    GOMP-ST: grasp optimized motion planning for suction transport,

    Y . Avigal, J. Ichnowski, M. Y . Cao, and K. Goldberg, “GOMP-ST: grasp optimized motion planning for suction transport,” inAlgorithmic Foundations of Robotics XV - Proceedings of the Fifteenth Workshop on the Algorithmic Foundations of Robotics, WAFR 2022, College Park, MD, USA, 22-24 June, 2022, ser. Springer Proceedings in Advanced Robotics, S. M. LaVal...

    work page 2022

  6. [6]

    Multi-Object Grasping in the Plane

    W. C. Agboh, J. Ichnowski, K. Goldberg, and M. R. Dogar, “Multi-object grasping in the plane,” 2022. [Online]. Available: https://arxiv.org/abs/2206.00229

    work page Pith review arXiv 2022

  7. [7]

    Learning to Efficiently Plan Robust Frictional Multi-Object Grasps

    W. C. Agboh, S. Sharma, K. Srinivas, M. Parulekar, G. Datta, T. Qiu, J. Ichnowski, E. Solowjow, M. Dogar, and K. Goldberg, “Learning to efficiently plan robust frictional multi-object grasps,” 2023. [Online]. Available: https://arxiv.org/abs/2210.07420

    work page Pith review arXiv 2023

  8. [8]

    Non-prehensile object transportation via model predictive non-sliding manipulation control,

    M. Selvaggio, A. Garg, F. Ruggiero, G. Oriolo, and B. Siciliano, “Non-prehensile object transportation via model predictive non-sliding manipulation control,”IEEE Transactions on Control Systems Tech- nology, pp. 1–14, 2023

    work page 2023

Show all 40 references
  1. [9]

    Keep it upright: Model predictive control for nonprehensile object transportation with obstacle avoidance on a mobile manipulator,

    A. Heins and A. P. Schoellig, “Keep it upright: Model predictive control for nonprehensile object transportation with obstacle avoidance on a mobile manipulator,”IEEE Robotics and Automation Letters, vol. 8, no. 12, p. 7986–7993, Dec. 2023. [Online]. Available: http://dx.doi.o...

    2023

  2. [10]

    Stable transport of assemblies: Pushing stacked parts,

    J. D. Bernheisel and K. M. Lynch, “Stable transport of assemblies: Pushing stacked parts,”IEEE Transactions on Automation science and Engineering, vol. 1, no. 2, pp. 163–168, 2004

    2004

  3. [11]

    Nonprehensile dynamic manipulation: A survey,

    F. Ruggiero, V . Lippiello, and B. Siciliano, “Nonprehensile dynamic manipulation: A survey,”IEEE Robotics and Automation Letters, vol. 3, no. 3, pp. 1711–1718, 2018

    2018

  4. [12]

    Nonprehensile manipulation: a trajectory-planning perspective,

    P. Acharya, K.-D. Nguyen, H. M. La, D. Liu, and I.-M. Chen, “Nonprehensile manipulation: a trajectory-planning perspective,” IEEE/ASME Transactions on Mechatronics, vol. 26, no. 1, pp. 527– 538, 2020

    2020

  5. [13]

    Topp-mpc- based dual-arm dynamic collaborative manipulation for multi-object nonprehensile transportation,

    C. Zhou, M. Lei, L. Zhao, Z. Wang, and Y . Zheng, “Topp-mpc- based dual-arm dynamic collaborative manipulation for multi-object nonprehensile transportation,” in2022 International Conference on Robotics and Automation (ICRA), 2022, pp. 999–1005

    2022

  6. [14]

    Swoosh! rattle! thump! – actions that sound,

    D. Gandhi, A. Gupta, and L. Pinto, “Swoosh! rattle! thump! – actions that sound,” 2020

    2020

  7. [15]

    That sounds right: Auditory self- supervision for dynamic robot manipulation,

    A. Thankaraj and L. Pinto, “That sounds right: Auditory self- supervision for dynamic robot manipulation,” inProceedings of The 7th Conference on Robot Learning, ser. Proceedings of Machine Learning Research, J. Tan, M. Toussaint, and K. Darvish, Eds., vol

  8. [16]

    Maniwav: Learning robot manipulation from in-the-wild audio-visual data,

    Z. Liu, C. Chi, E. Cousineau, N. Kuppuswamy, B. Burchfiel, and S. Song, “Maniwav: Learning robot manipulation from in-the-wild audio-visual data,”arXiv preprint arXiv:2406.19464, 2024

    2024 arXiv

  9. [17]

    Learning audio feedback for estimating amount and flow of granular material,

    S. Clarke, T. Rhodes, C. G. Atkeson, and O. Kroemer, “Learning audio feedback for estimating amount and flow of granular material,” inProceedings of The 2nd Conference on Robot Learning, ser. Proceedings of Machine Learning Research, A. Billard, A. Dragan, J. Peters, and J. Mo...

    2018

  10. [18]

    CHOMP: Gradient optimization techniques for efficient motion planning,

    N. Ratliff, M. Zucker, J. A. Bagnell, and S. Srinivasa, “CHOMP: Gradient optimization techniques for efficient motion planning,” in 2009 IEEE International Conference on Robotics and Automation. IEEE, 2009, pp. 489–494

    2009

  11. [19]

    STOMP: Stochastic trajectory optimization for motion planning,

    M. Kalakrishnan, S. Chitta, E. Theodorou, P. Pastor, and S. Schaal, “STOMP: Stochastic trajectory optimization for motion planning,” in2011 IEEE international conference on robotics and automation. IEEE, 2011, pp. 4569–4574

    2011

  12. [20]

    Finding locally optimal, collision-free trajectories with sequential convex optimization

    J. Schulman, J. Ho, A. X. Lee, I. Awwal, H. Bradlow, and P. Abbeel, “Finding locally optimal, collision-free trajectories with sequential convex optimization.” inRobotics: Science and Systems, 2013, pp. 1–10

    2013

  13. [21]

    GOMP: Grasp-optimized motion planning for bin picking,

    J. Ichnowski, M. Danielczuk, J. Xu, V . Satish, and K. Goldberg, “GOMP: Grasp-optimized motion planning for bin picking,” in2020 International Conference on Robotics and Automation (ICRA). IEEE, May 2020

    2020

  14. [22]

    Deep learning can accelerate grasp-optimized motion planning,

    J. Ichnowski, Y . Avigal, V . Satish, and K. Goldberg, “Deep learning can accelerate grasp-optimized motion planning,”Science Robotics, vol. 5, no. 48, 2020

    2020

  15. [23]

    Dynamic underactuated nonprehensile manipulation,

    K. M. Lynch and M. T. Mason, “Dynamic underactuated nonprehensile manipulation,” inProceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems. IROS’96, vol. 2. IEEE, 1996, pp. 889–896

    1996

  16. [24]

    Dynamic nonprehensile manipulation: Controllability, plan- ning, and experiments,

    ——, “Dynamic nonprehensile manipulation: Controllability, plan- ning, and experiments,”The International Journal of Robotics Re- search, vol. 18, no. 1, pp. 64–92, 1999

    1999

  17. [25]

    Using projected dynamics to plan dynamic contact manipulation,

    S. S. Srinivasa, M. A. Erdmann, and M. T. Mason, “Using projected dynamics to plan dynamic contact manipulation,” in2005 IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2005, pp. 3618–3623

    2005

  18. [26]

    Gomp-fit: Grasp- optimized motion planning for fast inertial transport,

    J. Ichnowski, Y . Avigal, Y . Liu, and K. Goldberg, “Gomp-fit: Grasp- optimized motion planning for fast inertial transport,” in2022 Inter- national Conference on Robotics and Automation (ICRA), 2022, pp. 5255–5261

    2022

  19. [27]

    Critically fast pick-and-place with suction cups,

    H. Pham and Q. Pham, “Critically fast pick-and-place with suction cups,”CoRR, vol. abs/1809.03151, 2018. [Online]. Available: http://arxiv.org/abs/1809.03151

    2018 arXiv

  20. [28]

    Tossing- bot: Learning to throw arbitrary objects with residual physics,

    A. Zeng, S. Song, J. Lee, A. Rodriguez, and T. Funkhouser, “Tossing- bot: Learning to throw arbitrary objects with residual physics,”IEEE Transactions on Robotics, vol. 36, no. 4, pp. 1307–1319, 2020

    2020

  21. [29]

    SwingBot: Learning Physical Features from In-hand Tactile Exploration for Dynamic Swing-up Manipulation,

    C. Wang, S. Wang, B. Romero, F. Veiga, and E. Adelson, “SwingBot: Learning Physical Features from In-hand Tactile Exploration for Dynamic Swing-up Manipulation,” 2020

    2020

  22. [30]

    Robots of the Lost Arc: Learning to Dynamically Manipulate Fixed-Endpoint Ropes and Cables,

    H. Zhang, J. Ichnowski, D. Seita, J. Wang, and K. Goldberg, “Robots of the Lost Arc: Learning to Dynamically Manipulate Fixed-Endpoint Ropes and Cables,” 2021

    2021

  23. [31]

    Iterative residual policy: for goal-conditioned dynamic manipulation of deformable objects,

    C. Chi, B. Burchfiel, E. Cousineau, S. Feng, and S. Song, “Iterative residual policy: for goal-conditioned dynamic manipulation of deformable objects,” 2022. [Online]. Available: https://arxiv.org/ abs/2203.00663

    2022 arXiv

  24. [32]

    curobo: Parallelized collision-free minimum-jerk robot motion generation,

    B. Sundaralingam, S. K. S. Hari, A. Fishman, C. Garrett, K. V . Wyk, V . Blukis, A. Millane, H. Oleynikova, A. Handa, F. Ramos, N. Ratliff, and D. Fox, “curobo: Parallelized collision-free minimum-jerk robot motion generation,” 2023

    2023

  25. [33]

    On-line computational scheme for mechanical manipulators,

    J. Y . Luh, M. W. Walker, and R. P. Paul, “On-line computational scheme for mechanical manipulators,” 1980

    1980

  26. [34]

    Jerk-limited real-time trajectory genera- tion with arbitrary target states,

    L. Berscheid and T. Kr ¨oger, “Jerk-limited real-time trajectory genera- tion with arbitrary target states,”Robotics: Science and Systems XVII, 2021

    2021

  27. [35]

    librosa: Audio and music signal analysis in python,

    B. McFee, C. Raffel, D. Liang, D. P. Ellis, M. McVicar, E. Battenberg, and O. Nieto, “librosa: Audio and music signal analysis in python,” inProceedings of the 14th python in science conference, 2015, pp. 18–25

    2015

  28. [36]

    SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python,

    P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Peterson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, ˙I. Polat, Y . Fen...

    2020

  29. [37]

    timsainb/noisereduce: v1.0,

    T. Sainburg, “timsainb/noisereduce: v1.0,” Jun. 2019. [Online]. Available: https://doi.org/10.5281/zenodo.3243139

    2019 doi

  30. [38]

    Finding, visualizing, and quantifying latent structure across diverse animal vocal repertoires,

    T. Sainburg, M. Thielk, and T. Q. Gentner, “Finding, visualizing, and quantifying latent structure across diverse animal vocal repertoires,” PLoS computational biology, vol. 16, no. 10, p. e1008228, 2020

    2020

  31. [39]

    Segment anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y . Lo, P. Doll ´ar, and R. Gir- shick, “Segment anything,”arXiv:2304.02643, 2023

    2023 arXiv

  32. [229]

    1036–1049

    PMLR, 06–09 Nov 2023, pp. 1036–1049. [Online]. Available: https://proceedings.mlr.press/v229/thankaraj23a.html

    2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.