Pith. sign in

REVIEW 3 major objections 4 minor 26 references

FLIP: Flowability-Informed Powder Weighing

T0 review · 3 major / 4 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 flowability-calibrated training pipeline cuts real-robot powder weighing error to 2.12 mg, beating a domain-randomized baseline by a factor of three.

desk verdict AoR-calibrated sims plus a flowability curriculum look genuinely better for robotic powder weighing, but the dynamic-fidelity claim is under-supported and the evaluation has selection and small-N issues. read the letter →

arxiv 2506.03896 v2 pith:W2TXOISH submitted 2025-06-04 cs.RO

Nikola Radulov , Alex Wright , Thomas Little , Andrew I. Cooper , Gabriella Pizzuto This is my paper
classification cs.RO
keywords powderweighingflowabilityangleofreposesim-to-realtransfercurriculumlearningBayesianoptimizationroboticmanipulationlaboratoryautomation
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 robotic powder weighing can be made reliable across very different powders by putting a measured physical property—flowability, captured by the angle of repose—explicitly into both the simulation and the training schedule. It claims that calibrating a physics simulator's parameters with Bayesian optimization until simulated piles match real powder piles produces training environments from which a policy transfers zero-shot to a physical robot. It further claims that ordering training from free-flowing to cohesive powders yields a 2.12 ± 1.53 mg average dispensing error, well below the 6.11 ± 3.92 mg of a domain-randomized baseline, and that the policy generalizes to cohesive powders and target masses it never saw. If true, flowability is a sufficient macroscopic descriptor to bridge the sim-to-real gap for granular material handling.

What carries the argument

The central object is the static angle of repose (AoR), the angle of the cone a powder forms when poured through a funnel; it is used as a stand-in for flowability, with higher AoR meaning less flowable powder. Around it FLIP builds a calibration loop: Bayesian optimization searches simulator parameters such as adhesion, cohesion, friction, particle size, and damping to minimize |AoR_real − AoR_sim(θ)|, retaining parameter sets within 1.5° of the real measurement. Those AoR-matched parameter sets define a flowability-conditioned training distribution, and a curriculum mechanism advances training through parameter sets in order of decreasing flowability once the policy meets an error threshold. The machinery carries the argument because it replaces broad parameter randomization with a physically meaningful scalar that groups powders by manipulation difficulty.

What would settle it

Measure a dynamic flow property, such as the mass dispensed per shake or the funnel discharge rate, for the five powders and for their AoR-matched simulations; if the simulated dynamic values differ from the real ones by more than the run-to-run noise for any powder the policy handles well, the static-AoR calibration premise is falsified.

Watch

Extended reading notes

Core claim

FLIP's central claim is that material flowability, quantified by the static angle of repose (AoR), is the right organizing variable for sim-to-real transfer in powder manipulation. The paper shows that automated AoR measurements (average error 0.84° vs. manual) can drive Bayesian optimization of a position-based-dynamics powder simulator, yielding parameter sets whose simulated static AoR matches five real powders within 0.22–1.37°. Training a soft actor-critic policy on these flowability-conditioned environments, ordered from lowest to highest AoR, produces a controller that dispenses powders with 2.12 ± 1.53 mg mean absolute error on a real robot, compared with 6.11 ± 3.92 mg for domain randomization, and that handles semolina and sodium bicarbonate—materials outside the 28°–37° training range—and a 20 mg target weight never seen in training.

Load-bearing premise

The whole transfer rests on one premise: a powder whose static angle of repose matches the simulation will also flow, shake, and dispense like the real powder during the dynamic weighing motion—but the paper only measures the static pile angle, never a dynamic flow property.

Editorial extensions

If this is right

  • A single macroscopic measurement, taken automatically in minutes, can calibrate a powder simulator well enough for zero-shot policy transfer.
  • Ordering training data by flowability matters: the curriculum beats both random and reverse-curriculum ordering, so the easy-to-hard sequence is itself a source of robustness.
  • The trained policy generalizes to out-of-distribution cohesive powders and to an untrained target mass, implying the learned controller is not tied to specific materials or setpoints.
  • All policy training happens in simulation, and the real robot inherits the policy without fine-tuning, making the pipeline practical for laboratory deployment.

Reading between the lines

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

  • The paper leaves implicit that the same Bayesian-calibration loop could be applied to other measurable macroscopic powder properties, such as dynamic discharge rate or compressibility, to build simulators for scooping, pouring, and tableting tasks.
  • The poor performance of the reverse curriculum hints that the ordering effect may be about learning stable corrective actions on easy powders before facing cohesive ones; a testable extension would vary curriculum thresholds and spacing to isolate that mechanism.
  • Because only static AoR is calibrated, the framework's guarantees are bounded by that descriptor; a natural stress test is to shift powder flowability by humidity and check whether re-measuring AoR and re-calibrating restores policy accuracy.
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 / 4 minor

Summary. The paper introduces FLIP, a framework for robotic powder weighing that uses the angle of repose (AoR) as a flowability metric to calibrate physics-based powder simulations via Bayesian optimization. The calibrated simulators are then used to train a reinforcement learning policy for the weighing task, with a curriculum that progressively increases material cohesiveness. The policy is transferred zero-shot to a real Franka robot and evaluated on five powders, including two with AoR values outside the training range. The authors report that the curriculum-based FLIP achieves a mean dispensing error of 2.12±1.53 mg, outperforming a domain-randomized baseline (6.11±3.92 mg) and other ablations.

Significance. If the central claim is upheld, the work is a meaningful contribution to sim-to-real transfer for granular material manipulation, combining a physically interpretable material descriptor (AoR) with simulation calibration and curriculum learning. The automated AoR measurement is validated against manual measurements, and the simulation calibration achieves low pile-geometry errors for the five tested powders. The paper also provides a useful comparison of curriculum orderings and a domain-randomization baseline. However, the significance is currently tempered by the lack of dynamic validation of the calibrated simulator and by evaluation choices that weaken the statistical strength of the real-world results.

major comments (3)
  1. [Section III-A.2 and IV-B] The simulation calibration and validation focus exclusively on the static angle of repose and final pile geometry (Table III), but the dispensing task itself is dynamic: the policy controls shaking and inclination to transfer a target mass, and the outcome depends on transient jamming, avalanching, and cohesion under motion. The paper presents no direct comparison of any dynamic flow property (e.g., discharged mass over time, discharge rate under the shake/incline protocol, or avalanche behavior) between simulation and reality. Since AoR is a single quasi-static scalar and many parameter combinations can reproduce it, the central claim that the AoR-calibrated simulator generates 'accurate training data' for the weighing task is not established. The authors should either measure a dynamic flow property in both domains or substantially temper the claim that AoR calibration is responsible for the sim-to-real transfer.
  2. [Section IV-D] The real-world evaluation protocol selects the best-performing simulation policy out of five seeds for each method before zero-shot transfer, and then each condition is evaluated with only five real trials. This double selection can bias the reported means, and no statistical significance test or confidence interval is provided for the headline comparison (FLIP(Curriculum) 2.12±1.53 mg vs Domain Randomisation 6.11±3.92 mg). The empirical claim would be much stronger if results were reported for all seeds (or a fixed selection rule were used) and if between-method differences were tested for significance.
  3. [Equation (2), Section III-A.1] The formula AoR = 2h/d_base is the tangent of the angle of repose, not the angle itself; the values in Table II are reported in degrees (e.g., 28.55°), which is inconsistent with the formula as written. If the implementation used arctan(2h/d_base), this should be stated explicitly; otherwise the calibration target is mis-specified and the numeric AoR values in the paper are not reproducible.
minor comments (4)
  1. [Algorithm 1 and Section IV-B] The acceptance threshold is stated as T_AoR = 1.5° in Section IV-B, but Algorithm 1 line 13 accepts a parameter set when E < 1; please reconcile these values.
  2. [Section IV-C] The paper states that the number of optimised data points per flowability level is set to 7, but it does not explain how these 7 parameter sets are chosen from the top 10 retained by the Bayesian optimization; please clarify the selection rule.
  3. [Throughout] There are several typographical errors, including 'signficantly' in the Introduction and 'maneouver' in Section IV-A.2; a careful proofread is recommended.
  4. [Conclusions] The phrase 'the first successful application of flowability-informed reinforcement learning' is too strong; consider softening to 'a demonstration' in light of the limited evidence and the issues raised above.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: AoR calibration and dispensing-error evaluation are distinct, and the OOD results come from direct physical experiments.

full rationale

The calibration loop only fits simulator micro-parameters to the static angle of repose (Algorithm 1, Eq. 1); the reported real-world outcome is dispensing error in milligrams on a physical robot, which is not a function of the fitted AoR by construction. The AoR residuals in Table III are the calibration objective itself, not a 'prediction', and the paper does not relabel them as such. The out-of-distribution powders (semolina, sodium bicarbonate) were excluded from the training flowability range [28, 37] degrees, and no policy parameter or curriculum threshold was fitted to their real dispensing errors; the reported 2.12 +/- 1.53 mg result is a direct physical measurement. Self-citations ([3], [4], [5], [14]) are contextual and none carries a load-bearing uniqueness or ansatz claim. The hand-tuned curriculum thresholds are transparently disclosed as hyperparameters tuned in simulation, not derived from the real evaluation data. The only substantive caveat, that AoR matching leaves dynamic flow properties unverified, is a validity and fidelity limitation explicitly acknowledged in Section III-A.2 and is not a circular reduction.

Assumptions & free parameters 12 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical entities. The central claims rest on 12 free parameters/hyperparameters: 9 simulation physics parameters fitted to real AoR data per powder, 2 curriculum thresholds and a points-per-level count, plus the chosen training flowability range. The AoR data is external and OOD powders provide an independent check, so circularity burden is low. The main assumptions are about simulation fidelity (spherical particles, PBD, static AoR as a proxy for dynamic difficulty), which are stated but not independently validated.

free parameters (12)
  • Particle Diameter = BO-optimized within 0.22-0.32 mm per powder
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Adhesion = BO-optimized within 0.0-1.3
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Particle Adhesion Scale = BO-optimized within 0.0-1.3
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Adhesion Offset Scale = BO-optimized within 0.0-0.00005
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Friction = BO-optimized within 0.0-1.3
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Gravity Scale = BO-optimized within 0.3-1.1
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Damping = BO-optimized within 0.0-1.0
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Cohesion = BO-optimized within 0.0-1.2
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Particle Mass = BO-optimized within 13-16.5 ug
    One of 9 simulation parameters fitted to reproduce real AoR; per-powder values not reported in the paper.
  • Curriculum thresholds = mean error < 0.8 mg, max error < 1 mg, 1330 episodes per level
    Manually tuned hyperparameters that control when the curriculum advances to a less flowable powder; no ablation is provided.
  • Optimized points per flowability level = 7
    Set by the authors to 'balance learning quality and stability'; the value was found empirically.
  • Training flowability range = 28-37 degrees AoR
    Chosen based on available real-world powders (sand, sugar, salt); excludes the more cohesive test powders.
assumptions (5)
  • domain assumption Position-Based Dynamics (PBD) in NVIDIA Isaac Sim provides sufficient physical fidelity for policy transfer.
    Section III-A.2 adopts PBD 'prioritising simulation speed at the expense of physical fidelity'; the method assumes this is accurate enough for training a dispensing policy.
  • domain assumption Spherical, uniform-size particles can represent real powder behavior.
    Section III-A.2 states particles are 'perfectly spherical' with 'identical, uniform size and mass,' neglecting polydispersity; without this, the simulator would be intractable.
  • domain assumption Angle of repose is a sufficient proxy for powder manipulation difficulty in dispensing.
    Section III-B.2 structures training around AoR as the key macroscopic property; if AoR does not track dynamic cohesion during shaking, the curriculum and calibration would not transfer.
  • domain assumption Simulation parameters calibrated to match static AoR also reproduce the dynamic dispensing behavior relevant to weighing.
    The optimization objective in Equation 1 is static AoR only; the evaluation is dynamic dispensing. This transfer is hypothesized, not directly measured.
  • domain assumption ISO 8398:1989 AoR measurement is a valid ground-truth procedure.
    Manual AoR values from this standard are used as calibration targets; the standard's validity is presumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FLIP: Flowability-Informed Powder Weighing." pith.science (2026). https://pith.science/paper/W2TXOISH

@misc{pith2026250603896,
  author       = {Pith},
  title        = {Pith review of: FLIP: Flowability-Informed Powder Weighing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W2TXOISH}},
  note         = {Machine review of arXiv:2506.03896}
}
read the original abstract

Autonomous manipulation of powders remains a significant challenge for robotic automation in scientific laboratories. The inherent variability and complex physical interactions of powders in flow, coupled with variability in laboratory conditions necessitates adaptive automation. This work introduces FLIP, a flowability-informed powder weighing framework designed to enhance robotic policy learning for granular material handling. Our key contribution lies in using material flowability, quantified by the angle of repose, to optimise physics-based simulations through Bayesian inference. This yields material-specific simulation environments capable of generating accurate training data, which reflects diverse powder behaviours, for training "robot chemists". Building on this, FLIP integrates quantified flowability into a curriculum learning strategy, fostering efficient acquisition of robust robotic policies by gradually introducing more challenging, less flowable powders. We validate the efficacy of our method on a robotic powder weighing task under real-world laboratory conditions. Experimental results show that FLIP with a curriculum strategy achieves a low dispensing error of 2.12 +/- 1.53 mg, outperforming methods that do not leverage flowability data, such as domain randomisation (6.11 +/- 3.92 mg). These results demonstrate FLIP's improved ability to generalise to previously unseen, more cohesive powders and to new target masses.

Figures

Figures reproduced from arXiv: 2506.03896 by the authors.

Figure 1
Figure 1. The FLIP framework leverages BO to calibrate simulation parameters using real-world powder flowability data. This flowability￾informed powder simulation then serves as a high-fidelity environment for training the robot control policy for powder weighing. validated against manual methods. 3) A simulation calibration framework that leverages measured material flowability data to improve granular material dynamics with… view at source ↗
Figure 2
Figure 2. Measurement of static powder flowability using the AoR [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Total reward (a) and final powder weighing error (b), [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

26 extracted references · 21 canonical work pages

  1. [1]

    Self-driving laboratories for chemistry and materials science,

    G. Tom, S. P. Schmid, S. G. Baird, Y . Cao, K. Darvish, H. Hao, S. Lo, S. Pablo-Garc´ıa, E. M. Rajaonson, M. Skreta, and et al., “Self-driving laboratories for chemistry and materials science,”Chemical Reviews, 2024

    work page 2024

  2. [2]

    A mobile robotic chemist,

    B. Burger, P. M. Maffettone, V . V . Gusev, C. M. Aitchison, Y . Bai, X. yan Wang, X. Li, B. M. Alston, B. Li, R. Clowes, N. Rankin, B. Harris, R. S. Sprick, and A. I. Cooper, “A mobile robotic chemist,”Nature, vol. 583, pp. 237 – 241, 2020. [Online]. Available: https://api.semanticscholar.org/CorpusID:220420261

    work page 2020

  3. [3]

    Modular, multi-robot integration of laboratories: an autonomous workflow for solid-state chemistry,

    A. M. Lunt, H. Fakhruldeen, G. Pizzuto, L. Longley, A. White, N. Rankin, R. Clowes, B. Alston, L. Gigli, G. M. Day, A. I. Cooper, and S. Y . Chong, “Modular, multi-robot integration of laboratories: an autonomous workflow for solid-state chemistry,” Chem. Sci., vol. 15, pp. 2456–2463, 2024. [Online]. Available: http://dx.doi.org/10.1039/D3SC06206F

    work page doi:10.1039/d3sc06206f 2024

  4. [4]

    Autonomous mobile robots for exploratory synthetic chemistry,

    T. Dai, S. Vijayakrishnan, F. T. Szczypi ´nski, J.-F. Ayme, E. Simaei, T. Fellowes, R. Clowes, L. Kotopanov, C. E. Shields, Z. Zhou, J. W. Ward, and A. I. Cooper, “Autonomous mobile robots for exploratory synthetic chemistry,”Nature, Nov 2024. [Online]. Available: https://doi.org/10.1038/s41586-024-08173-7

    doi:10.1038/s41586-024-08173-7 2024

  5. [5]

    Autonomous biomimetic solid dispensing using a dual-arm robotic manipulator,

    Y . Jiang, H. Fakhruldeen, G. Pizzuto, L. Longley, A. He, T. Dai, R. Clowes, N. Rankin, and A. I. Cooper, “Autonomous biomimetic solid dispensing using a dual-arm robotic manipulator,”Digital Discovery, vol. 2, pp. 1733–1744, 2023. [Online]. Available: http://dx.doi.org/10.1039/D3DD00075C

    work page doi:10.1039/d3dd00075c 2023

  6. [6]

    Learning robotic pow- der weighing from simulation for laboratory automation,

    Y . Kadokawa, M. Hamaya, and K. Tanaka, “Learning robotic pow- der weighing from simulation for laboratory automation,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, pp. 2932–2939

    work page 2023

  7. [7]

    On powder flowability,

    J. Prescott and R. Barnum, “On powder flowability,”Pharmaceutical Technology, vol. 24, pp. 60–84+236, 01 2000

    work page 2000

  8. [8]

    Curriculum learning,

    Y . Bengio, J. Louradour, R. Collobert, and J. Weston, “Curriculum learning,” inProceedings of the 26th Annual International Conference on Machine Learning, ser. ICML ’09. New York, NY , USA: Association for Computing Machinery, 2009, p. 41–48. [Online]. Available: https://doi.org/10.1145/1553374.1553380

    arXiv 2009

Show all 26 references
  1. [9]

    A multi-robot–multi-task scheduling system for autonomous chemistry laboratories,

    J. Zhou, M. Luo, L. Chen, Q. Zhu, S. Jiang, F. Zhang, W. Shang, and J. Jiang, “A multi-robot–multi-task scheduling system for autonomous chemistry laboratories,”Digital Discovery, pp. –, 2025. [Online]. Available: http://dx.doi.org/10.1039/D4DD00313F

    2025 doi

  2. [10]

    Organa: A robotic assistant for automated chemistry experimentation and characterization,

    K. Darvish, M. Skreta, Y . Zhao, N. Yoshikawa, S. Som, M. Bogdanovic, Y . Cao, H. Hao, H. Xu, A. Aspuru-Guzik, A. Garg, and F. Shkurti, “Organa: A robotic assistant for automated chemistry experimentation and characterization,”Matter, vol. 8, no. 2, p. 101897, 2025. [Online]. ...

    2025

  3. [11]

    Leveraging multi-modal sensing for robotic insertion tasks in r&d laboratories,

    A. Butterworth, G. Pizzuto, L. Pecyna, A. I. Cooper, and S. Luo, “Leveraging multi-modal sensing for robotic insertion tasks in r&d laboratories,” in2023 IEEE 19th International Conference on Automa- tion Science and Engineering (CASE), 2023, pp. 1–8

    2023

  4. [12]

    High-accuracy injection using a mobile manipulation robot for chemistry lab automation,

    A. Angelopoulos, M. Verber, C. McKinney, J. Cahoon, and R. Al- terovitz, “High-accuracy injection using a mobile manipulation robot for chemistry lab automation,” in2023 IEEE/RSJ International Con- ference on Intelligent Robots and Systems (IROS), 2023, pp. 10 102– 10 109

    2023

  5. [13]

    Reliable and robust robotic handling of microplates via computer vision and touch feedback,

    V . Scamarcio, J. Tan, F. Stellacci, and J. Hughes, “Reliable and robust robotic handling of microplates via computer vision and touch feedback,”Frontiers in Robotics and AI, vol. 11, 2025. [Online]. Available: https://www.frontiersin.org/ journals/robotics-and-ai/articles/10....

    2025

  6. [14]

    Accelerating laboratory automation through robot skill learning for sample scraping,

    G. Pizzuto, H. Wang, H. Fakhruldeen, B. Peng, K. S. Luck, and A. I. Cooper, “Accelerating laboratory automation through robot skill learning for sample scraping,” 2022

    2022

  7. [15]

    Robotic powder grinding with audio-visual feedback for laboratory automation in materials science,

    Y . Nakajima, M. Hamaya, K. Tanaka, T. Hawai, F. von Drigalski, Y . Takeichi, Y . Ushiku, and K. Ono, “Robotic powder grinding with audio-visual feedback for laboratory automation in materials science,” 10 2023

    2023

  8. [16]

    Domain randomization for transferring deep neural networks from simulation to the real world,

    J. Tobin, R. Fong, A. Ray, J. Schneider, W. Zaremba, and P. Abbeel, “Domain randomization for transferring deep neural networks from simulation to the real world,” 2017. [Online]. Available: https://arxiv.org/abs/1703.06907

    2017 arXiv

  9. [17]

    Sim-to-real transfer for visual reinforcement learning of deformable object manipulation for robot- assisted surgery,

    P. M. Scheikl, E. Tagliabue, B. Gyenes, M. Wagner, D. Dall’Alba, P. Fiorini, and F. Mathis-Ullrich, “Sim-to-real transfer for visual reinforcement learning of deformable object manipulation for robot- assisted surgery,”IEEE Robotics and Automation Letters, vol. 8, no. 2, pp. 5...

    2023

  10. [18]

    Inferring the material properties of granular media for robotic tasks,

    C. Matl, Y . S. Narang, R. Bajcsy, F. Ramos, and D. Fox, “Inferring the material properties of granular media for robotic tasks,”2020 IEEE International Conference on Robotics and Automation (ICRA), pp. 2770–2777, 2020. [Online]. Available: https://api.semanticscholar.org/Corp...

    2020

  11. [19]

    Stir to pour: Efficient calibration of liquid properties for pouring actions,

    T. Lopez-Guevara, R. Pucci, N. K. Taylor, M. U. Gutmann, S. Ra- mamoorthy, and K. Suhr, “Stir to pour: Efficient calibration of liquid properties for pouring actions,” in2020 IEEE/RSJ International Con- ference on Intelligent Robots and Systems (IROS), 2020, pp. 5351– 5357

    2020

  12. [20]

    Decoding powder flowability: Machine learning pioneers the analysis of particle-size distribution effects,

    W. Liu, Z. Deng, Y . Zhang, X. Zhu, J. Huang, H. Zhang, and J. Zhu, “Decoding powder flowability: Machine learning pioneers the analysis of particle-size distribution effects,”Powder Technology, vol. 435, p. 119407, 2024. [Online]. Available: https: //www.sciencedirect.com/sci...

    2024

  13. [21]

    Practical bayesian opti- mization of machine learning algorithms,

    J. Snoek, H. Larochelle, and R. P. Adams, “Practical bayesian opti- mization of machine learning algorithms,” inProceedings of the 26th International Conference on Neural Information Processing Systems - V olume 2, ser. NIPS’12. Red Hook, NY , USA: Curran Associates Inc., 2012...

    2012

  14. [22]

    Position based dynamics,

    M. M ¨uller, B. Heidelberger, M. Hennix, and J. Ratcliff, “Position based dynamics,”J. Vis. Comun. Image Represent., vol. 18, no. 2, p. 109–118, Apr. 2007. [Online]. Available: https://doi.org/10.1016/j. jvcir.2007.01.005

    2007 doi

  15. [23]

    NVIDIA Isaac Sim,

    NVIDIA, “NVIDIA Isaac Sim,” NVIDIA, 2025. [Online]. Available: https://developer.nvidia.com/isaac-sim

    2025

  16. [24]

    R. S. Sutton and A. G. Barto,Reinforcement Learning: An Introduc- tion. Cambridge, MA, USA: A Bradford Book, 2018

    2018

  17. [25]

    Orbit: A unified simulation framework for interactive robot learning environments,

    M. Mittal, C. Yu, Q. Yu, J. Liu, N. Rudin, D. Hoeller, J. L. Yuan, P. P. Tehrani, R. Singh, Y . Guo, H. Mazhar, A. Mandlekar, B. Babich, G. State, M. Hutter, and A. Garg, “Orbit: A unified simulation framework for interactive robot learning environments,” 2023

    2023

  18. [26]

    Soft actor-critic: Off- policy maximum entropy deep reinforcement learning with a stochastic actor,

    T. Haarnoja, A. Zhou, P. Abbeel, and S. Levine, “Soft actor-critic: Off- policy maximum entropy deep reinforcement learning with a stochastic actor,” inICML, 2018

    2018

Pith tools

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