Pith. sign in

REVIEW 3 major objections 6 minor 51 references

Data efficient Robotic Object Throwing with Model-Based Reinforcement Learning

T0 review · 3 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read A model-based RL algorithm, MC-PILOT, learns to throw objects to nearly every target from a single exploratory trial.

desk verdict Competent, incremental MBRL paper for robot throwing, but the data-efficiency claim against Model-Free RL is not supported by the experiments because the baseline is supervised regression, not RL. read the letter →

arxiv 2502.05595 v1 pith:KVCBBI3H submitted 2025-02-08 cs.RO

Niccolò Turcato , Giulio Giacomuzzo , Matteo Terreran , Davide Allegro , Ruggero Carli , Alberto Dalla Libera This is my paper · ORCID
classification cs.RO MSC 68T4068T05
keywords pick-and-throwmodel-basedreinforcementlearningGaussianprocessregressionreleasedelayestimationdataefficiencypolicygeneralizationFrankaEmikaPandaMonteCarlooptimization
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

MC-PILOT is a model-based reinforcement learning method for pick-and-throw: a robot grasps an object and throws it into a target bin rather than carrying it there. The paper's central claim is that by fitting a Gaussian-process model of the object's free flight and an estimated release-delay distribution, the algorithm learns a throwing policy that hits nearly every target in simulation and on a Franka Emika Panda arm using only a handful of exploratory throws. The policy maps any target position in a region to a release speed, so new target locations are hit without collecting new data. Compared with an analytic ballistic baseline and with model-free policies trained on growing datasets, MC-PILOT reaches higher accuracy with far less interaction. If this holds, it makes pick-and-throw practical for industrial cells where targets change frequently and interaction time is expensive.

What carries the argument

The central object is the particle-based rollout with a delayed-release initial state. Each particle samples a target $P^{(m)}$, a release delay $t_d^{(m)}\sim U(a,a+b)$, and starts from $x_0^{(m)} = [f_{\mathrm{kin}}(q_{\tilde t_r}^{(m)}), J_a(q_{\tilde t_r}^{(m)})\dot q_{\tilde t_r}^{(m)}]^T$, then propagates through one-step GP Gaussian transitions. This makes the policy optimization aware of both model uncertainty and timing uncertainty. A second mechanism is the delay estimator: Bayesian optimization minimizes the distance between GP-simulated landing points and measured landings as a function of $(a,b)$, using the same GP model. The speed-integration GP model with a squared-exponential kernel carries the free-flight dynamics.

What would settle it

In simulation the true delay is uniform on $[0.01, 0.02]$ s, but the estimated distribution is $U(0.021, 0.024)$ s, already showing the estimator is not recovering the physical delay; on the real arm, a high-speed camera or instrumented gripper that records the actual separation event across velocities and objects would settle whether one uniform scalar delay exists, and if the fitted $(a,b)$ must change with object and speed to keep landings accurate, the release model is a calibration term rather than the stated mechanism.

Watch

Extended reading notes

Core claim

MC-PILOT inserts a GP model of the free-flight dynamics and a uniform release-delay distribution $t_d \sim U(a,a+b)$ into the initial condition of each Monte Carlo particle; the policy is a squashed radial-basis network $\pi_\theta(P)$ that outputs release speed from target $P$, and optimization propagates particles through the GP to minimize expected saturated landing distance. In simulation with $N_{\mathrm{exp}}=5$, $N_a=0$ and one outer-loop trial, the paper reports almost 100% accuracy on batches of 100 sampled targets, whereas the ballistic baseline misses the far targets and a neural-network policy needs many more training throws. On the real Franka Emika Panda with $N_{\mathrm{exp}}=10$, $N_a=2$, a single trial yields policies that hit training and unseen test targets within a 0.1 m radius, and the bin-target experiment reaches 100% accuracy with three objects while the baseline reaches roughly 50%. The release-delay distribution is estimated by Bayesian optimization so that GP-simulated landings match observed landings, and it is re-estimated per object.

Load-bearing premise

The whole scheme assumes that the object lets go at a single random instant and that the robot's real motion at that instant matches the planned motion, so the release position and velocity can be taken straight from the plan.

Editorial extensions

If this is right

  • A single trial of about ten exploration throws suffices to hit a target region, so pick-and-throw becomes usable in settings where hundreds of practice throws are not affordable.
  • Because the policy is a function of target position, generalization to unseen targets follows from the same trial without additional exploration.
  • Modeling the release delay inside the particle simulation is the difference between near-100% accuracy and baseline-like misses; ignoring it leaves far targets unreachable.
  • When task requirements change, such as replacing ground targets with a bin of different height, the same model supports a new policy by re-running optimization only, in about 15 minutes on a laptop GPU.
  • Delay parameters are object-dependent and must be re-estimated when the thrown object changes, while the GP model itself can be reused across objects.

Reading between the lines

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

  • The estimated $(a,b)$ likely functions as a lumped calibration that absorbs tracking error, drag, and release mechanics rather than as a literal physical delay; a natural test is to compare the fitted distribution against high-speed video of actual object separation.
  • The fixed-release-geometry assumption, a constant height, angle, and radial distance, is what makes the policy a scalar speed from target; extending to variable release angle or height would stress whether the GP can absorb those extra degrees of freedom.
  • The same delay-estimation loop could be repurposed for other latency-sensitive dynamic manipulation skills, such as catching, hitting, or handover, where the arm and end-effector are not synchronized.
  • The rotational data augmentation and target-augmented state exploit symmetry around the robot's vertical axis; deployment in asymmetric environments, such as wind or uneven ground, would require revisiting those choices.
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 / 6 minor

Summary. The paper introduces MC-PILOT, a model-based reinforcement learning method for robotic pick-and-throw with a Franka Emika Panda. It learns a Gaussian process model of free-flight dynamics from motion-capture trajectories, estimates a uniform release-delay distribution via Bayesian optimization on landing predictions, and optimizes a squashed radial-basis-function policy by Monte Carlo particle simulation. Experiments in Gazebo and on the real system report near-100% target accuracy from a single trial (with Nexp=10 in the real setup) and generalization to unseen target locations and to several object shapes, comparing against a ballistic baseline and a neural-network regression baseline.

Significance. If the empirical claims hold, this is a valuable demonstration that model-based policy search can solve a dynamic manipulation task with very little real-world interaction. The paper's concrete strengths are the real-hardware validation with multiple object shapes, the explicit treatment of the gripper release delay inside the policy optimization loop, and the use of GP predictive uncertainty in a particle-based policy gradient scheme. However, the headline comparison against 'Model-Free RL' is not supported by the experiments as reported, because the baseline is a supervised regression network rather than a reinforcement learning algorithm. This limits the force of the data-efficiency claim, although the absolute throwing accuracy achieved by MC-PILOT is reported consistently across simulation and hardware.

major comments (3)
  1. [§6.2, §6.3.3, §7] The 'Model-Free' baseline is not a Model-Free RL algorithm. In §6.2 the neural-network policy is trained by minimizing the mean squared error between applied velocities and landing positions on collected throws, with no reward signal, no policy-gradient update, and no exploration loop. Section 7 nevertheless states that MC-PILOT was compared with a 'Model-Free RL solution' and concludes that it is more data-efficient. This mismatch is load-bearing because the central claim is data efficiency relative to Model-Free RL. Please either add a genuine MFRL baseline (e.g., PPO or DDPG with the same policy class and reward) or restrict the comparative claims to the supervised regression baseline and the analytical baseline actually evaluated.
  2. [§6.2 and Table 2 (left)] The simulation validation of the delay estimator is not convincing. The true release delay is U(0.01, 0.02) s, while the estimated distribution is a = 0.021 ± 0.057 s and b = 0.003 ± 0.004 s. The estimated interval [0.021, 0.024] does not match the true support, and the large standard deviation on a means that 'the optimization does predict the presence of a delay statistically' is not supported by the numbers. Since the paper presents delay estimation as a separate contribution, please report a calibration or coverage measure of the estimated distribution against the true delay, or explicitly discuss what quantity the estimator is capturing if it is not the physical release delay.
  3. [§3.2, eq. (24)] The release model assumes that the robot tracks the reference trajectory closely enough that the release state is obtained by forward kinematics at the sampled release time, but no tracking-error quantification is provided, despite Figure 4 showing actual and reference joint trajectories. The simulation result above suggests that the fitted (a, b) parameters may be absorbing model mismatch and tracking error rather than representing a physical delay. To support the delay-estimation contribution, please quantify the tracking error at release and analyze how it affects the estimated delay distribution.
minor comments (6)
  1. [§5] The delay distribution is defined inconsistently: §5 first says td ∼ U(a, b), but eq. (29) and the surrounding text use td ∼ U(a, a + b). Please choose one convention and use it throughout.
  2. [§6.2] The true simulation delay is written with a comma as the decimal separator ('[0,01 s, 0,02 s]'); use consistent decimal notation and notation matching Table 2.
  3. [§6.3.3] The neural-network baseline is called 'unsupervised' in the text, but it is trained by supervised regression on collected landing and velocity data. Rename it to avoid confusion.
  4. [§4.2.2] The simulation-time variable is t′ in §4 but t in the update rule in eq. (25); unify the notation.
  5. [References] References [14] and [31] are the same paper; the duplicate should be removed.
  6. [Figure 4] The caption says 'actual trajectory recorded on the robot,' but the text does not discuss the magnitude of tracking error; either add a quantitative comparison or remove the word 'actual' if the curves are only references.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; central throwing-accuracy claims are externally validated on the real robot and on held-out targets.

full rationale

MC-PILOT's performance claims are validated against external benchmarks rather than being constructed from its own inputs. The policy is optimized on a GP dynamics model and an estimated release-delay distribution, but the reported accuracy is measured on the real Franka Emika Panda (Section 6.3.3, Fig. 11) and in simulation on 100-target batches across 10 seeds (Section 6.2), including held-out test and demo targets. The delay parameters (a,b) are obtained in Section 5 by minimizing, via Bayesian optimization, the mismatch between GP-simulated and measured landing positions (eq. 29); this is a calibration to observed landings, and the paper does not present (a,b) as an independent prediction. The final throwing success is not forced by this fit because real-system landings are used for evaluation. Self-citations to MC-PILCO [19] and to the preliminary paper [21] supply algorithmic components (GP speed-integration model, dropout, RBF policy), but the manuscript states the model and optimization equations explicitly, and no central claim relies solely on an unverified self-citation. The weak comparison baseline (supervised neural-network regression described as 'Model-Free') is a correctness/validation concern, not a circularity under the rubric. No ansatz is smuggled in via citation: the uniform delay model is stated as an assumption and fitted to data.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central algorithm rests on a learned GP dynamics model, a fitted delay distribution, and several tuning choices; no new physical entities are introduced. The main additional assumptions are the instantaneous scalar release-delay model and the sufficiency of COM-state dynamics.

free parameters (4)
  • GP kernel hyperparameters (signal variance, lengthscales, noise variance) = not reported
    Trained per output dimension by maximizing marginal likelihood (Section 4.1, eqs. (2)-(3)); the model accuracy depends on these values.
  • Release delay distribution parameters (a,b) = object-dependent; rubber ball a=0.259s b=0.009s, sim a=0.021s b=0.003s (Table 2)
    Fitted by Bayesian optimization in Section 5 to minimize simulated vs actual landing distance; central to delay-aware policy optimization.
  • Initial gripper opening delay compensation = 0.24s
    Chosen by hand from a landing-distance experiment (Section 6.3.1) and used for the Baseline and MC-PILOT(tr=tr) variants.
  • Algorithm hyperparameters (Nexp, Na, Nopt, M, Md, Nb, uM, Ts, T, lc) = see Table 1 (e.g., Nexp=10, M=400, Nb=250, lc=0.1m)
    Chosen by the authors for simulation and real setup; performance depends on these choices, especially exploration trials Nexp and cost length lc.
assumptions (5)
  • standard math Gaussian process regression posterior equations and squared-exponential kernel are standard results.
    Used in Section 4.1 to derive one-step-ahead dynamics from eqs. (2)-(3), relying on [32].
  • domain assumption Object free flight is represented by center-of-mass position and velocity, with acceleration changes modeled as independent GPs.
    Section 4.1, eq. (18); neglects orientation, spin, and non-COM dynamics, acknowledged for cylinder and hammer.
  • domain assumption Release is an instantaneous event with a scalar uniform delay td, and the robot tracks its reference trajectory closely enough that the release state equals forward kinematics at trcmd + td.
    Section 3.2 eqs. (11)-(12), Section 4.2.2 eq. (24); load-bearing for particle initialization.
  • ad hoc to paper Random rotation of collected trajectories around the vertical axis preserves throwing dynamics.
    Algorithm 1, Section 4.1; relies on rotational symmetry of the setup and approximately symmetric objects.
  • ad hoc to paper A squashed radial-basis-function network can represent the optimal target-to-velocity mapping.
    Section 4.2.1, eq. (21); chosen function class, not derived from first principles.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Data efficient Robotic Object Throwing with Model-Based Reinforcement Learning." pith.science (2026). https://pith.science/paper/KVCBBI3H

@misc{pith2026250205595,
  author       = {Pith},
  title        = {Pith review of: Data efficient Robotic Object Throwing with Model-Based Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KVCBBI3H}},
  note         = {Machine review of arXiv:2502.05595}
}
read the original abstract

Pick-and-place (PnP) operations, featuring object grasping and trajectory planning, are fundamental in industrial robotics applications. Despite many advancements in the field, PnP is limited by workspace constraints, reducing flexibility. Pick-and-throw (PnT) is a promising alternative where the robot throws objects to target locations, leveraging extrinsic resources like gravity to improve efficiency and expand the workspace. However, PnT execution is complex, requiring precise coordination of high-speed movements and object dynamics. Solutions to the PnT problem are categorized into analytical and learning-based approaches. Analytical methods focus on system modeling and trajectory generation but are time-consuming and offer limited generalization. Learning-based solutions, in particular Model-Free Reinforcement Learning (MFRL), offer automation and adaptability but require extensive interaction time. This paper introduces a Model-Based Reinforcement Learning (MBRL) framework, MC-PILOT, which combines data-driven modeling with policy optimization for efficient and accurate PnT tasks. MC-PILOT accounts for model uncertainties and release errors, demonstrating superior performance in simulations and real-world tests with a Franka Emika Panda manipulator. The proposed approach generalizes rapidly to new targets, offering advantages over analytical and Model-Free methods.

Figures

Figures reproduced from arXiv: 2502.05595 by the authors.

Figure 1
Figure 1. Panda Robot executing the throwing task with target bin. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Screenshots from the simulation, robot in release configuration. The [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. The robot performs a throwing motion. 4. MC-PILOT This section describes the MBRL algorithm developed to train the robot for the object-throwing task. The algorithm is called MC-PILOT and is derived from MC-PILCO [19]. The pseudocode of MC-PILOT is reported in Algorithm 1. MC￾PILOT is a Model-Based policy gradient algorithm, in which GPR is used to estimate system dynamics, and long-term state distributions are appr… view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Reference joint trajectory and actual trajectory recorded on the robot, for the tossing motion. The trajectory moves only three joints. Plot shows the [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Results of the Neural Network policy trained on datasets of increasing size. The target positions projected on the horizontal plane are colored green if the [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: (a): Plots of the target reach accuracy of 10 batches of trials for each [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: The tested objects. On the front row, from left: the tennis ball, and the [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Delay compensation estimation experiment with the rubber ball. For [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Top: landing distances predicted by the the simulations with the GP [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 11
Figure 11. Figure 11: Accuracy tests with the rubber ball. The numbered circles are placed [PITH_FULL_IMAGE:figures/full_fig_p014_11.png]
Figure 11
Figure 11. Figure 11: fig. 11. The results are presented both in the form of 2D plots, [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 14
Figure 14. Figure 14: Accuracy tests with the cube. The numbered red circles are test [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: Accuracy tests with the tennis ball. The numbered red circles are [PITH_FULL_IMAGE:figures/full_fig_p016_15.png]
Figure 16
Figure 16. Figure 16: Accuracy tests with the cylinder (left) and the hammer (right). Each [PITH_FULL_IMAGE:figures/full_fig_p017_16.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 38 canonical work pages

  1. [1]

    R. A. Brooks, Planning collision- free motions for pick-and-place op- erations, The International Journal of Robotics Research 2 (4) (1983) 19–44. arXiv:https://doi.org/10.1177/027836498300200402, doi:10.1177/027836498300200402. URL https://doi.org/10.1177/027836498300200402

    work page doi:10.1177/027836498300200402 1983

  2. [2]

    Zhang, B

    Y . Zhang, B. K. Chen, X. Liu, Y . Sun, Autonomous robotic pick-and- place of microobjects, IEEE Transactions on Robotics 26 (1) (2010) 200–

    work page 2010

  3. [3]

    N. C. Dafle, A. Rodriguez, R. Paolini, B. Tang, S. S. Srinivasa, M. Erd- mann, M. T. Mason, I. Lundberg, H. Staab, T. Fuhlbrigge, Extrinsic dex- terity: In-hand manipulation with external forces, in: IEEE ICRA, 2014, pp. 1578–1585. doi:10.1109/ICRA.2014.6907062

    arXiv 2014

  4. [4]

    T. Chen, J. Xu, P. Agrawal, A system for general in-hand object re- orientation, in: A. Faust, D. Hsu, G. Neumann (Eds.), Proceedings of the 5th Conference on Robot Learning, V ol. 164 of Proceedings of Ma- chine Learning Research, PMLR, 2022, pp. 297–307. URL https://proceedings.mlr.press/v164/chen22a.html

    work page 2022

  5. [5]

    Raptopoulos, M

    F. Raptopoulos, M. Koskinopoulou, M. Maniadakis, Robotic pick-and- toss facilitates urban waste sorting, in: 2020 IEEE 16th International Conference on Automation Science and Engineering (CASE), 2020, pp. 1149–1154. doi:10.1109/CASE48305.2020.9216746

    arXiv 2020

  6. [6]

    LeCun, Y

    Y . LeCun, Y . Bengio, G. Hinton, Deep learning, nature 521 (7553) (2015) 436–444

    2015

  7. [7]

    C ¸ alıs ¸ır, M

    S. C ¸ alıs ¸ır, M. K. Pehlivano˘glu, Model-free reinforcement learning algo- rithms: A survey, in: 2019 27th signal processing and communications applications conference (SIU), IEEE, 2019, pp. 1–4

    work page 2019

  8. [8]

    Ghadirzadeh, A

    A. Ghadirzadeh, A. Maki, D. Kragic, M. Bj ¨orkman, Deep predictive pol- icy training using reinforcement learning, in: 2017 IEEE /RSJ Interna- tional Conference on Intelligent Robots and Systems (IROS), 2017, pp. 2351–2358. doi:10.1109/IROS.2017.8206046

    work page arXiv 2017

Show all 51 references
  1. [9]

    Pinsler, R

    R. Pinsler, R. Akrour, T. Osa, J. Peters, G. Neumann, Sample and feed- back e fficient hierarchical reinforcement learning from human prefer- ences, in: 2018 IEEE International Conference on Robotics and Automa- tion (ICRA), 2018, pp. 596–601. doi:10.1109/ICRA.2018.8460907

    2018

  2. [10]

    Kober, A

    J. Kober, A. Wilhelm, E. Oztop, J. Peters, Reinforcement learning to ad- just parametrized motor primitives to new situations, Autonomous Robots 33 (2012) 361–379

    2012

  3. [11]

    Gutzeit, A

    L. Gutzeit, A. Fabisch, M. Otto, J. H. Metzen, J. Hansen, F. Kirchner, E. A. Kirchner, The besman learning platform for automated robot skill learning, Frontiers in Robotics and AI 5 (2018) 43

    2018

  4. [12]

    Pahi ˇc, Z

    R. Pahi ˇc, Z. Lonˇcarevi´c, A. Gams, A. Ude, Robot skill learning in latent space of a deep autoencoder neural network, Robotics and Autonomous Systems 135 (2021) 103690. doi:https://doi.org/10.1016/j. robot.2020.103690

    2021

  5. [13]

    Hu, M.-C

    J.-S. Hu, M.-C. Chien, Y .-J. Chang, S.-H. Su, C.-Y . Kai, A ball-throwing robot with visual feedback, in: 2010 IEEE /RSJ International Conference on Intelligent Robots and Systems, 2010, pp. 2511–2512.doi:10.1109/ IROS.2010.5649335. 17

    2010

  6. [14]

    Monastirsky, O

    M. Monastirsky, O. Azulay, A. Sintov, Learning to throw with a handful of samples using decision transformers, IEEE Robotics and Automation Letters 8 (2) (2023) 576–583. doi:10.1109/LRA.2022.3229266

    2023

  7. [15]

    Huang, Y

    B. Huang, Y . Chen, T. Wang, Y . Qin, Y . Yang, N. Atanasov, X. Wang, Dy- namic handover: Throw and catch with bimanual hands, in: Conference on Robot Learning, PMLR, 2023, pp. 1887–1902

    2023

  8. [16]

    A. Zeng, S. Song, J. Lee, A. Rodriguez, T. Funkhouser, Tossingbot: Learning to throw arbitrary objects with residual physics, IEEE Transac- tions on Robotics 36 (4) (2020) 1307–1319. doi:10.1109/TRO.2020. 2988642

    2020 doi

  9. [17]

    Z. Fang, Y . Hou, J. Li, A pick-and-throw method for enhancing robotic sorting ability via deep reinforcement learning, in: 2021 36th Youth Aca- demic Annual Conference of Chinese Association of Automation (Y AC), 2021, pp. 479–484. doi:10.1109/YAC53711.2021.9486466

    2021

  10. [18]

    T. M. Moerland, J. Broekens, A. Plaat, C. M. Jonker, et al., Model-based reinforcement learning: A survey, Foundations and Trends® in Machine Learning 16 (1) (2023) 1–118

    2023

  11. [19]

    Amadio, A

    F. Amadio, A. Dalla Libera, R. Antonello, D. Nikovski, R. Carli, D. Romeres, Model-based policy search using monte carlo gradient es- timation with real systems application, IEEE Transactions on Robotics 38 (6) (2022) 3879–3898. doi:10.1109/TRO.2022.3184837

    2022

  12. [20]

    Amadio, A

    F. Amadio, A. Dalla Libera, D. Nikovski, R. Carli, D. Romeres, Learn- ing control from raw position measurements, in: 2023 American Control Conference (ACC), IEEE, 2023, pp. 2171–2178

    2023

  13. [21]

    Turcato, A

    N. Turcato, A. Dalla Libera, G. Giacomuzzo, R. Carli, Teaching a robot to toss arbitrary objects with model-based reinforcement learning, in: 2023 9th International Conference on Control, Decision and Information Tech- nologies (CoDIT), 2023, pp. 1126–1131. doi:10.1109/CoDIT58...

    2023

  14. [22]

    ai olympics with realaigym

    F. Wiebe, N. Turcato, A. Dalla Libera, C. Zhang, T. Vincent, S. Vyas, G. Giacomuzzo, R. Carli, D. Romeres, A. Sathuluri, et al., Reinforce- ment learning for athletic intelligence: Lessons from the 1st “ai olympics with realaigym” competition,”, in: Proceedings of the Thirty-T...

    2024

  15. [23]

    Taylor, A

    O. Taylor, A. Rodriguez, Optimal shape and motion planning for dynamic planar manipulation, Autonomous Robots 43 (2019) 327–344

    2019

  16. [24]

    Sintov, A

    A. Sintov, A. Shapiro, A stochastic dynamic motion planning algorithm for object-throwing, in: 2015 IEEE International Conference on Robotics and Automation (ICRA), 2015, pp. 2475–2480. doi:10.1109/ICRA. 2015.7139530

    2015

  17. [25]

    Paraschos, C

    A. Paraschos, C. Daniel, J. R. Peters, G. Neumann, Probabilistic move- ment primitives, Advances in neural information processing systems 26 (2013)

    2013

  18. [26]

    V . Mnih, K. Kavukcuoglu, D. Silver, A. A. Rusu, J. Veness, M. G. Bellemare, A. Graves, M. Riedmiller, A. K. Fidjeland, G. Ostrovski, et al., Human-level control through deep reinforcement learning, nature 518 (7540) (2015) 529–533

    2015

  19. [27]

    T. P. Lillicrap, J. J. Hunt, A. Pritzel, N. Heess, T. Erez, Y . Tassa, D. Silver, D. Wierstra, Continuous control with deep reinforcement learning, arXiv preprint arXiv:1509.02971 (2015)

    2015 arXiv

  20. [28]

    Bianchi, M

    D. Bianchi, M. G. Antonelli, C. Laschi, A. M. Sabatini, E. Falotico, Soft- oss: Learning to throw objects with a soft robot, IEEE Robotics & Au- tomation Magazine (2023) 2–12doi:10.1109/MRA.2023.3310865

    2023

  21. [29]

    Della Santina, M

    C. Della Santina, M. G. Catalano, A. Bicchi, M. Ang, O. Khatib, B. Si- ciliano, Soft robots, Encyclopedia of Robotics 489 (2020)

    2020

  22. [30]

    Schulman, F

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, O. Klimov, Proximal policy optimization algorithms, arXiv preprint arXiv:1707.06347 (2017)

    2017 arXiv

  23. [31]

    Monastirsky, O

    M. Monastirsky, O. Azulay, A. Sintov, Learning to throw with a handful of samples using decision transformers, IEEE Robotics and Automation Letters 8 (2) (2022) 576–583

    2022

  24. [32]

    C. E. Rasmussen, Gaussian processes in machine learning, in: Summer school on machine learning, Springer, 2003, pp. 63–71

    2003

  25. [33]

    Car `e, R

    A. Car `e, R. Carli, A. D. Libera, D. Romeres, G. Pillonetto, Kernel methods and gaussian processes for system identification and control: A road map on regularized kernel-based learning for control, IEEE Control Systems Magazine 43 (5) (2023) 69–110. doi:10.1109/MCS.2023. 3291625

    2023 doi

  26. [34]

    R. E. Caflisch, Monte carlo and quasi-monte carlo methods, Acta numer- ica 7 (1998) 1–49

    1998

  27. [35]

    Bottou, Large-scale machine learning with stochastic gradient descent, in: Proc of COMPSTAT’2010, Springer, 2010, pp

    L. Bottou, Large-scale machine learning with stochastic gradient descent, in: Proc of COMPSTAT’2010, Springer, 2010, pp. 177–186

    2010

  28. [36]

    D. P. Kingma, M. Welling, Auto-encoding variational bayes, arXiv preprint arXiv:1312.6114 (2013)

    2013 arXiv

  29. [37]

    D. P. Kingma, J. Ba, Adam: A method for stochastic optimization, arXiv preprint arXiv:1412.6980 (2014)

    2014 arXiv

  30. [38]

    Srivastava, G

    N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, R. Salakhutdinov, Dropout: a simple way to prevent neural networks from overfitting, JMLR 15 (1) (2014) 1929–1958

    2014

  31. [39]

    P. I. Frazier, A tutorial on bayesian optimization, arXiv preprint arXiv:1807.02811 (2018)

    2018 arXiv

  32. [40]

    Shahriari, K

    B. Shahriari, K. Swersky, Z. Wang, R. P. Adams, N. de Freitas, Taking the human out of the loop: A review of bayesian optimization, Proceed- ings of the IEEE 104 (1) (2016) 148–175. doi:10.1109/JPROC.2015. 2494218

    2016 doi

  33. [41]

    Evangelista, D

    D. Evangelista, D. Allegro, M. Terreran, A. Pretto, S. Ghidoni, An uni- fied iterative hand-eye calibration method for eye-on-base and eye-in- hand setups, in: 2022 IEEE 27th International Conference on Emerg- ing Technologies and Factory Automation (ETFA), 2022, pp. 1–7. doi:...

    2022

  34. [42]

    Allegro, M

    D. Allegro, M. Terreran, S. Ghidoni, Multi-camera hand-eye calibra- tion for human-robot collaboration in industrial robotic workcells (2024). arXiv:2406.11392

    2024 arXiv

  35. [43]

    Paszke, S

    A. Paszke, S. Gross, S. Chintala, G. Chanan, E. Yang, Z. DeVito, Z. Lin, A. Desmaison, L. Antiga, A. Lerer, Automatic di fferentiation in pytorch (2017)

    2017

  36. [44]

    Nogueira, Bayesian Optimization: Open source constrained global optimization tool for Python (2014–)

    F. Nogueira, Bayesian Optimization: Open source constrained global optimization tool for Python (2014–). URL https://github.com/bayesian-optimization/ BayesianOptimization

    2014

  37. [45]

    Koenig, A

    N. Koenig, A. Howard, Design and use paradigms for gazebo, an open- source multi-robot simulator, in: 2004 IEEE /RSJ IROS, V ol. 3, IEEE, 2004, pp. 2149–2154

    2004

  38. [46]

    Koub ˆaa, et al., Robot Operating System (ROS)., V ol

    A. Koub ˆaa, et al., Robot Operating System (ROS)., V ol. 1, Springer, 2017

    2017

  39. [47]

    Almedeij, Drag coe fficient of flow around a sphere: Matching asymp- totically the wide trend, Powder Technology 186 (3) (2008) 218–223

    J. Almedeij, Drag coe fficient of flow around a sphere: Matching asymp- totically the wide trend, Powder Technology 186 (3) (2008) 218–223

    2008

  40. [48]

    Coleman, I

    D. Coleman, I. Sucan, S. Chitta, N. Correll, Reducing the barrier to en- try of complex robotic software: a moveit! case study, arXiv preprint arXiv:1404.3785 (2014)

    2014 arXiv

  41. [49]

    Chitta, E

    S. Chitta, E. Marder-Eppstein, W. Meeussen, V . Pradeep, A. Rodr ´ıguez Tsouroukdissian, J. Bohren, D. Coleman, B. Magyar, G. Raiola, M. L¨udtke, E. Fern´andez Perdomo, ros control: A generic and simple control framework for ros, J. Open Source Softw. (2017)

    2017

  42. [50]

    Olson, Apriltag: A robust and flexible visual fiducial system, in: 2011 IEEE International Conference on Robotics and Automation, 2011, pp

    E. Olson, Apriltag: A robust and flexible visual fiducial system, in: 2011 IEEE International Conference on Robotics and Automation, 2011, pp. 3400–3407. doi:10.1109/ICRA.2011.5979561. 18

    2011

  43. [207]

    doi:10.1109/TRO.2009.2034831

    2009

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.