Pith. sign in

REVIEW 3 major objections 5 minor 79 references

Robust Contact-rich Manipulation through Implicit Motor Adaptation

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Policies retrieved by optimizing a parameter-weighted advantage function outperform policies conditioned on a single estimated parameter in contact-rich manipulation, given a roughly known parameter distribution.

desk verdict Solid empirical system paper, but the central theoretical claim does not hold; reviewers should focus on the experiments and demand a corrected or dropped proof. read the letter →

arxiv 2412.11829 v3 pith:Y4K2BZHE submitted 2024-12-16 cs.RO

Teng Xue , Amirreza Razmjoo , Suhan Shetty , Sylvain Calinon This is my paper · ORCID
classification cs.RO
keywords policyadaptationcontact-richmanipulationparametersimplicitmotorparameter-conditioned
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The robot learns a base policy in simulation for a manipulation skill such as pushing. Instead of storing the policy as a neural network that maps observations directly to actions, the policy is stored as a scoring function: for every combination of physical parameters, robot state, and candidate action, it scores how good that action is. The method uses tensor trains, a compact mathematical format, to store this scoring function. At run time, a small neural network looks at the robot's recent motion and produces a range of likely physical parameters, for example a rough range for mass, rather than one exact value. The robot then averages the scores over the entire range of parameters and takes the action with the highest average score. This is different from older explicit motor adaptation, which feeds one estimated parameter into the policy. Averaging scores before choosing an action can avoid committing to a bad single estimate. The paper tests hitting, pushing, and reorienting objects with different masses, sizes, and friction coefficients. In simulation, this implicit method gives lower final state error than explicit adaptation when the estimate is rough, and the authors demonstrate pushing a bleach bottle with a real Franka robot. The practical claim is plausible, but the mathematical proof that implicit adaptation is always better is essentially definitional, because it chooses the action that maximizes the very average used for comparison. The width of the parameter range is also chosen separately for each task after looking at the estimator error, so the robustness depends on a hand-set hyperparameter.
Extended reading notes

Core claim

IMA estimates the true parameter alpha using a probability distribution alphaHat ~ P(alphaHat) and retrieves the parameter-conditioned policy through domain contraction, with the inequality J_pi_i(x, P(alphaHat)) >= J_pi_e(x, P(alphaHat)) stated in Section 5.4, Eq. (27). If the paper is correct, a single simulation-trained, tensor-train-represented advantage function can produce robust contact-rich manipulation policies for new instances using only a rough probabilistic parameter estimate, without precise system identification or student-policy retraining.

Load-bearing premise

The load-bearing premise is that deployment-time parameter uncertainty is well captured by a uniform distribution U(nu_t - w/2, nu_t + w/2) whose center comes from an MLP and whose width w is chosen per task. If the MLP is biased by more than w/2, the domain-contracted advantage function averages over the wrong parameter instances and the claimed robustness, including Eq. (27), loses its grounding. The paper selects w after inspecting estimator error: N/20 for Hit, N/5 for Push and Reorientation, with N/3 favored in Fig. 10. The central result therefore depends on an evaluation-informed modeling choice rather than a prior derived from the theory.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes implicit motor adaptation (IMA), a framework for contact-rich manipulation under uncertain physical parameters. A parameter-augmented base policy is learned offline in tensor-train (TT) form as an implicit representation of the advantage function. At deployment, a probabilistic estimate of the system parameters (a uniform distribution with mean from an MLP and bandwidth w) is used to retrieve a parameter-conditioned policy via "domain contraction," i.e., by taking the argmax over the probability-weighted sum of parameter-specific advantage functions. The authors claim that IMA is theoretically superior to explicit motor adaptation (EMA), which feeds a single point estimate into the base policy. The paper reports simulations on Hit, Push, and Reorientation primitives and a real-robot planar pushing experiment, showing that IMA outperforms EMA and domain randomization, with efficient core-level TT operations for policy retrieval.

Significance. If the practical claims hold, the paper offers a useful engineering contribution: IMA avoids precise system identification and student-policy retraining, and the TT representation makes probabilistic policy retrieval fast enough for real-time control. The simulation protocol includes multiple randomized instances with error bars, and the real-robot demonstration with disturbances is a valuable stress test. However, the central theoretical proposition on which the paper's optimality claims rest is not correct as stated. The empirical comparison to EMA is still informative and appears to support the weaker, practical claim that IMA works well with rough estimates in these tasks, but the theoretical superiority argument in Proposition 2 is unsound and needs to be replaced or substantially reframed.

major comments (3)
  1. [§5.3, Eq. (15)] Equation (15) defines V(x|P(α̂)) as the expectation of the parameter-specific value functions V(x|α̂_j). This is not the value function of the deployment problem in Eq. (24): a single policy π must act without knowing α, both at the current state and at all future states, so its value under the distribution satisfies V_P^π(x)=E_α[R(x,π(x))+γ V_P^π(f_α(x,π(x)))], not Σ_j P_j V_{α_j}(x). Consequently, the representation A(x,u|P)=Σ_j P_j A_{α_j}(x,u) in Eq. (13) does not follow from Eqs. (14)–(18), and the argmax in Eq. (19) is not generally optimal for Eq. (24). A simple counterexample shows the failure: with equiprobable α1 and α2 and γ=1, from state x0 action a leads to x1 for both parameters while action b leads to x2 for both; at x1 one action gives reward 100 under α1 and 0 under α2, and another gives 0 under α1 and 110 under α2; at x2 any action gives 60. Domain contraction evaluates a as (100+110)/2=105 and b as 60, so it selects a, but any single deployment policy that reaches x1 must choose one action for both parameters, giving at most max(50,55)=55, while b yields 60. Thus the proposed contraction is not the solution of the stated robust-control objective.
  2. [§5.4, Eq. (27)] The inequality J_πi(x,P(α̂)) ≥ J_πe(x,P(α̂)) is asserted in Eq. (27) as a consequence of the previous definitions, but it is not proven. Since the IMA objective in Eq. (26) relies on the same incorrect V(x|P(α̂)) definition identified above, the comparison to EMA does not establish optimality. Moreover, the statement that "α̂ can be considered a single sample from P(α̂)" does not by itself imply that the EMA policy's value is bounded by the IMA policy's value; such an inequality requires additional assumptions on the value function (e.g., concavity or smoothness in α) and on the closed-loop behavior of the shared base policy. The empirical results in Section 6.3 are an independent and useful evaluation of the practical claim, but Proposition 2 as stated should be withdrawn or replaced by a proof under explicit and testable assumptions.
  3. [§6.2, §6.3, Fig. 10] The bandwidth w of the uniform estimator is a free hyperparameter that is selected after inspecting estimator errors: w=N/20 for Hit, w=N/5 for Push and Reorientation, and Fig. 10 indicates that w=N/3 gives the best results under the achieved estimation accuracy. The robustness claim that IMA works with rough estimates depends on the assumption that the true parameter lies inside the support U(ν_t−w/2, ν_t+w/2); because w is tuned using knowledge of the estimator's error on the evaluation tasks, the experiments do not demonstrate performance under a fixed, a priori uncertainty model. Please report results on held-out instances with w fixed before observing estimator errors, or provide a principled rule for setting w from the estimator's noise covariance rather than from the same test data used for evaluation.
minor comments (5)
  1. [§5.1, Eq. (9)] In Eq. (9), the policy in the Bellman operator is written as π_k(x), but the surrounding text and Eq. (8) define the policy as a function of both α and x; the notation should be π_k(α,x) for consistency.
  2. [§5.2] The MLP in the probabilistic adaptation module outputs a point estimate ν_t, while Eq. (12) writes a distribution P(α̂_t)=ϕ(·) directly; please clarify that the distribution is assumed uniform with mean ν_t and fixed width w, and state how w is chosen for the estimator.
  3. [§5.3, Eqs. (22)–(23)] The factorization P(j1,...,jd)=P_1(α̂_j1)···P_d(α̂_jd) assumes independence across parameter dimensions; if the estimator produces correlated parameter beliefs, the core-level contraction in Eq. (23) no longer applies, so this assumption should be stated explicitly.
  4. [§6.5, Table 3] The table header appears to repeat "Task" and the reported error/time values are not aligned with the column names as described in the text; please reformat the table and add clear units and definitions of the error metric.
  5. [Throughout] There are minor typographical errors, including "emplyed" in Section 2.2 and "arbitary" in Section 5.3, and the MSE loss in Section 5.2 should be written with a squared norm, i.e., MSE(ν_t, α_t)=∥ν_t−α_t∥².

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; Proposition 2 has a formal inference gap but no reduction by construction.

full rationale

Walking the derivation chain, I find no step in which a claimed prediction or first-principles result reduces to its inputs by construction. The base-policy machinery (TTPI, TTGO) is cited from prior work (Shetty et al. 2024a,b), but those are published, independently reusable algorithms and the paper's empirical comparison against SAC-based RL+DR and RL+EMA provides external support for the practical claims. The uniform-bandwidth hyperparameter w is chosen with awareness of the estimator error (Fig. 10, with w=N/3 favored there and w=N/20 or N/5 used in the main comparison), but this is an explicit, transparent hyperparameter selection, not a fitted quantity renamed as a prediction. The only serious issue is Proposition 2 in Section 5.4: the step from Eqs. (25)-(26) to Eq. (27) is not proved, because V(x|P(α̂)) in Eq. (26) is a weighted average of parameter-specific value functions and is not shown to equal the actual objective Jπ in Eq. (24); the 'therefore' is a formal gap. That is a correctness or inference problem, not a circular reduction: inequality (27) is asserted rather than being an identity forced by the definitions. The self-citations to domain contraction (Xue et al. 2024a) are descriptive and not load-bearing for the validity of the empirical comparisons. Hence the circularity score is 0.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The free-parameter count is modest but consequential: the bandwidth w changes the effective domain and is tuned per task after error analysis, and reward shaping defines what optimal means. The axioms are standard modeling assumptions plus one ad hoc assumption about the uniform support. The empirical comparison is therefore not derivable from first principles; it depends on several hand-set choices.

free parameters (2)
  • uniform bandwidth w = N/20 for Hit, N/5 for Push and Reorientation; N/3 favored in Fig. 10
    Section 5.2 defines P(alphaHat) = U(nu_t - w/2, nu_t + w/2). The value is task-specific and informed by the estimator error analysis, so it behaves as a fitted hyperparameter rather than a predicted one.
  • reward shaping weights = 0.01 action and velocity penalties, beta=10^4 for Reorientation, l_p=0.005, l_o=0.01*pi
    Eqs. (28) through (32) define the optimality criterion for each task. Hand-chosen weights influence which policy counts as optimal, so all comparisons depend on these choices.
assumptions (4)
  • ad hoc to paper The true physical parameter lies inside the estimated uniform support U(nu_t - w/2, nu_t + w/2).
    Section 5.2 and Eq. (12): if the MLP estimate is biased by more than w/2, domain contraction averages over advantage functions that exclude the true dynamics, and the robustness claim in Eq. (27) has no force.
  • domain assumption TT-Cross and TTGO give sufficiently accurate value and advantage functions and near-global policy retrieval.
    Section 5.1 relies on Shetty et al. 2024a and 2024b; no convergence or error bound for the TTPI value iteration is proved in this paper.
  • domain assumption The simulation dynamics for push and reorientation match the real robot sufficiently.
    Section 6.1 uses quasi-static limit-surface pushing and a torsional friction model based on Vina et al. 2016; real-world success depends on this modeling premise.
  • domain assumption The MLP parameter estimator trained with MSE provides a mean nu_t that is not systematically biased.
    Section 5.2: only the MSE loss is minimized; no calibration or bias guarantee is provided for the estimated mean.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Robust Contact-rich Manipulation through Implicit Motor Adaptation." pith.science (2026). https://pith.science/paper/Y4K2BZHE

@misc{pith2026241211829,
  author       = {Pith},
  title        = {Pith review of: Robust Contact-rich Manipulation through Implicit Motor Adaptation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y4K2BZHE}},
  note         = {Machine review of arXiv:2412.11829}
}
read the original abstract

Contact-rich manipulation plays an important role in daily human activities. However, uncertain physical parameters often pose significant challenges for both planning and control. A promising strategy is to develop policies that are robust across a wide range of parameters. Domain adaptation and domain randomization are widely used, but they tend to either limit generalization to new instances or perform conservatively due to neglecting instance-specific information. \textit{Explicit motor adaptation} addresses these issues by estimating system parameters online and then retrieving the parameter-conditioned policy from a parameter-augmented base policy. However, it typically requires precise system identification or additional training of a student policy, both of which are challenging in contact-rich manipulation tasks with diverse physical parameters. In this work, we propose \textit{implicit motor adaptation}, which enables parameter-conditioned policy retrieval given a roughly estimated parameter distribution instead of a single estimate. We leverage tensor train as an implicit representation of the base policy, facilitating efficient retrieval of the parameter-conditioned policy by exploiting the separable structure of tensor cores. This framework eliminates the need for precise system estimation and policy retraining while preserving optimal behavior and strong generalization. We provide a theoretical analysis to validate the approach, supported by numerical evaluations on three contact-rich manipulation primitives. Both simulation and real-world experiments demonstrate its ability to generate robust policies across diverse instances. Project website: \href{https://sites.google.com/view/implicit-ma}{https://sites.google.com/view/implicit-ma}.

Figures

Figures reproduced from arXiv: 2412.11829 by the authors.

Figure 1
Figure 1. Deployment of the learned policy in a variety of contact-rich manipulation tasks. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. We can access the element indexed (i1, . . . , id) of the tensor in this format simply by multiplying matrix slices from the cores: F(i1,...,id) = F 1 :,i1,:F 2 :,i2,: · · · F d :,id,: , (2) where F k :,ik,: ∈ R rk−1×rk represents the ik-th frontal slice (a matrix) of the third-order tensor F k . For any given [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Pipeline of the proposed approach, including (1) parameter-augmented base policy learning, (2) probabilistic system adaptation with proprioceptive history, and (3) parameter-conditioned policy retrieval. The base policy and parameter-conditioned policy are implicitly represented by the corresponding advantage functions A(α, x, u) and Aˆ(x, u), respectively. Blue-shaded modules are trained in simulation, and green-sh… view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Domain contraction in TT format. The parameter-augmented advantage function in TT format typically includes separate 3rd-order cores for different dimensionality, such as parameter, state and action. Given a probabilistic parameter distribution, we can retrieve the par…
Figure 5
Figure 5. Figure 5: Domain contraction unifies domain randomization and domain adaptation by giving different parameter distributions. estimate it as αˆ, drawn from a probability distribution P(αˆ). This distribution can be inferred from the proprioceptive history, namely P(αˆt) = ϕ(xt−k:…
Figure 6
Figure 6. Figure 6: Position and velocity responses of a spring-damper system, along with the approximated value functions (figure insets as colormaps), for the policies derived from the ground truth, EMA, and IMA. This indicates that the policy obtained by EMA typically demonstrates poor…
Figure 7
Figure 7. Figure 7: Comparison of final state error given different estimated parameter distribution [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Comparison between IMA and EMA on three manipulation primitives in terms of the final state error, given the same base policy [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Resulting hitting impact of the policies derived from the ground truth, EMA, and IMA. generally outperforms EMA on the three contact-rich manipulation tasks, particularly in the open-loop hitting task. In this paper, we employ a simple MLP with a structure of 512 × 256…
Figure 12
Figure 12. Figure 12: Trajectories produced by IMA (Left) and EMA (Right) for the reorientation task. The blue lines depict the trajectory of an arbitrary object, with the objective of reorienting it from the bottom configuration to a perfectly vertical top configuration. The lengths of th…
Figure 11
Figure 11. Figure 11: Planar pushing tasks with a sugar box (Top) and a mustard bottle (Bottom). Left: Object trajectory produced by IMA; Right: Object trajectory produced by EMA. 2021), using a Multilayer Perceptron (MLP) with a 64 × 64 × 64 architecture as the policy network. The discoun…
Figure 13
Figure 13. Figure 13: Comparison of different robust primitive learning approaches in terms of training time, and state errors of base policies and adapted policies. The unit of training time is seconds for Hit and minutes for Push and Reorientation. Errors are calculated based on the ℓ2 n…
Figure 14
Figure 14. Figure 14: An example of planar pushing task: bleach bottle on a metal surface. The system begins in the initial configuration (a), aiming to manipulate the bottle to reach the target configuration (h). The robot first makes contact with the bottle and pushes it slightly (b), th…
Figure 15
Figure 15. Figure 15: Sensitivity analysis of policy performance with respect to the maximum TT-rank (solid lines) and discretization granularity (dashed lines). The y-axis shows the normalized ℓ2 error between the final and target states, scaled by the worst-case performance in each task.…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

79 extracted references · 59 canonical work pages

  1. [1]

    Robust adaptive mpc for constrained uncertain nonlinear systems

    Veronica Adetola and Martin Guay. Robust adaptive mpc for constrained uncertain nonlinear systems. International Journal of Adaptive Control and Signal Processing, 25 0 (2): 0 155--167, 2011

    work page 2011

  2. [2]

    Adaptive model predictive control for constrained nonlinear systems

    Veronica Adetola, Darryl DeHaan, and Martin Guay. Adaptive model predictive control for constrained nonlinear systems. Systems & Control Letters, 58 0 (5): 0 320--326, 2009

    work page 2009

  3. [3]

    Meta reinforcement learning for sim-to-real domain adaptation

    Karol Arndt, Murtaza Hazara, Ali Ghadirzadeh, and Ville Kyrki. Meta reinforcement learning for sim-to-real domain adaptation. In Proc. IEEE Intl Conf. on Robotics and Automation ( ICRA ) , pages 2725--2731, 2020

    work page 2020

  4. [4]

    Simultaneous learning of contact and continuous dynamics

    Bibit Bianchini, Mathew Halm, and Michael Posa. Simultaneous learning of contact and continuous dynamics. In Conference on Robot Learning, pages 3966--3978. PMLR, 2023

    work page 2023

  5. [5]

    Robot programming by demonstration

    Aude Billard, Sylvain Calinon, Ruedige Dillmann, and Stefan Schaal. Robot programming by demonstration. In B. Siciliano and O. Khatib, editors, Handbook of Robotics, pages 1371--1394. Springer, Secaucus, NJ, USA, 2008

    work page 2008

  6. [6]

    Using simulation and domain adaptation to improve efficiency of deep robotic grasping

    Konstantinos Bousmalis, Alex Irpan, Paul Wohlhart, Yunfei Bai, Matthew Kelcey, Mrinal Kalakrishnan, Laura Downs, Julian Ibarz, Peter Pastor, Kurt Konolige, Sergey Levine, and Vincent Vanhoucke. Using simulation and domain adaptation to improve efficiency of deep robotic grasping. In Proc. IEEE Intl Conf. on Robotics and Automation ( ICRA ) , pages 4243--4...

    work page 2018

  7. [7]

    The YCB object and model set: Towards common benchmarks for manipulation research

    Berk Calli, Arjun Singh, Aaron Walsman, Siddhartha Srinivasa, Pieter Abbeel, and Aaron M Dollar. The YCB object and model set: Towards common benchmarks for manipulation research. In 2015 international conference on advanced robotics (ICAR), pages 510--517. IEEE, 2015

    work page 2015

  8. [8]

    Closing the sim-to-real loop: Adapting simulation randomization with real world experience

    Yevgen Chebotar, Ankur Handa, Viktor Makoviychuk, Miles Macklin, Jan Issac, Nathan Ratliff, and Dieter Fox. Closing the sim-to-real loop: Adapting simulation randomization with real world experience. In Proc. IEEE Intl Conf. on Robotics and Automation ( ICRA ) , pages 8973--8979, 2019

    work page 2019

Show all 79 references
  1. [9]

    League: Guided skill learning and abstraction for long-horizon manipulation

    Shuo Cheng and Danfei Xu. League: Guided skill learning and abstraction for long-horizon manipulation. IEEE Robotics and Automation Letters ( RA-L ) , 8 0 (10): 0 6451--6458, 2023

    2023

  2. [10]

    Diffusion policy: Visuomotor policy learning via action diffusion

    Cheng Chi, Zhenjia Xu, Siyuan Feng, Eric Cousineau, Yilun Du, Benjamin Burchfiel, Russ Tedrake, and Shuran Song. Diffusion policy: Visuomotor policy learning via action diffusion. International Journal of Robotics Research (IJRR), 0 0 (0), 2024

    2024

  3. [11]

    Reinforcement learning with vision-proprioception model for robot planar pushing

    Lin Cong, Hongzhuo Liang, Philipp Ruppel, Yunlei Shi, Michael G \"o rner, Norman Hendrich, and Jianwei Zhang. Reinforcement learning with vision-proprioception model for robot planar pushing. Frontiers in Neurorobotics, 16: 0 829437, 2022

    2022

  4. [12]

    Data-driven tensor train gradient cross approximation for hamilton--jacobi--bellman equations

    Sergey Dolgov, Dante Kalise, and Luca Saluzzi. Data-driven tensor train gradient cross approximation for hamilton--jacobi--bellman equations. SIAM Journal on Scientific Computing, 45 0 (5): 0 A2153--A2184, 2023

    2023

  5. [13]

    Palm-e: An embodied multimodal language model

    Danny Driess, Fei Xia, Mehdi SM Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, Wenlong Huang, Chebotar Yevgen, Pierre Sermanet, Daniel Duckworth, Sergey Levine, Vincent Vanhoucke, Karol Hausman, Marc Toussaint, K...

    2023

  6. [14]

    Implicit generation and modeling with energy based models

    Yilun Du and Igor Mordatch. Implicit generation and modeling with energy based models. In Advances in Neural Information Processing Systems ( NIPS ) , volume 32, 2019

    2019

  7. [15]

    Sliding mode control: theory and applications

    Christopher Edwards and Sarah Spurgeon. Sliding mode control: theory and applications. CRC Press, 1998

    1998

  8. [16]

    Implicit behavioral cloning

    Pete Florence, Corey Lynch, Andy Zeng, Oscar A Ramirez, Ayzaan Wahid, Laura Downs, Adrian Wong, Johnny Lee, Igor Mordatch, and Jonathan Tompson. Implicit behavioral cloning. In Conference on Robot Learning, pages 158--168. PMLR, 2022

    2022

  9. [17]

    Robust nonlinear control associating robust feedback linearization and H_ control

    Ana L \`u cia D Franco, Henri Bourles, Edson R De Pieri, and Herve Guillard. Robust nonlinear control associating robust feedback linearization and H_ control. IEEE Transactions on Automatic Control, 51 0 (7): 0 1200--1207, 2006

    2006

  10. [18]

    Physically grounded vision-language models for robotic manipulation

    Jensen Gao, Bidipta Sarkar, Fei Xia, Ted Xiao, Jiajun Wu, Brian Ichter, Anirudha Majumdar, and Dorsa Sadigh. Physically grounded vision-language models for robotic manipulation. In Proc. IEEE Intl Conf. on Robotics and Automation ( ICRA ) , pages 12462--12469. IEEE, 2024

    2024

  11. [19]

    Model predictive control: Theory and practice—a survey

    Carlos E Garcia, David M Prett, and Manfred Morari. Model predictive control: Theory and practice—a survey. Automatica, 25 0 (3): 0 335--348, 1989

    1989

  12. [20]

    PDDL The Planning Domain Definition Language

    Malik Ghallab, Adele Howe, Craig Knoblock, Drew McDermott, Ashwin Ram, Manuela Veloso, Daniel Weld, David Wilkins, Anthony Barrett, Dave Christianson, Marc Friedman, Chung Kwok, Keith Golden, Scott Penberthy, David E Smith, Ying Sun, and Daniel Weld. PDDL The Planning Domain D...

    1998

  13. [21]

    Reinforcement learning with deep energy-based policies

    Tuomas Haarnoja, Haoran Tang, Pieter Abbeel, and Sergey Levine. Reinforcement learning with deep energy-based policies. In International conference on machine learning, pages 1352--1361. PMLR, 2017

    2017

  14. [22]

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

    Tuomas Haarnoja, Aurick Zhou, Pieter Abbeel, and Sergey Levine. Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor. In International conference on machine learning, pages 1861--1870. PMLR, 2018

    2018

  15. [23]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 0 6840--6851, 2020

    2020

  16. [24]

    Reactive planar non-prehensile manipulation with hybrid model predictive control

    Francois R Hogan and Alberto Rodriguez. Reactive planar non-prehensile manipulation with hybrid model predictive control. International Journal of Robotics Research (IJRR), 39 0 (7): 0 755--773, 2020 a

    2020

  17. [25]

    Feedback control of the pusher-slider system: A story of hybrid and underactuated contact dynamics

    Fran c ois Robert Hogan and Alberto Rodriguez. Feedback control of the pusher-slider system: A story of hybrid and underactuated contact dynamics. In Algorithmic Foundations of Robotics XII: Proceedings of the Twelfth Workshop on the Algorithmic Foundations of Robotics, pages ...

    2020

  18. [26]

    Reinforcement learning: A survey

    Leslie Pack Kaelbling, Michael L Littman, and Andrew W Moore. Reinforcement learning: A survey. Journal of artificial intelligence research, 4: 0 237--285, 1996

    1996

  19. [27]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In Proc.\ Intl Conf.\ on Learning Representations ( ICLR ) , 2015

    2015

  20. [28]

    Reinforcement learning in robotics: A survey

    Jens Kober, J Andrew Bagnell, and Jan Peters. Reinforcement learning in robotics: A survey. The International Journal of Robotics Research, 32 0 (11): 0 1238--1274, 2013

    2013

  21. [29]

    RMA : Rapid motor adaptation for legged robots

    Ashish Kumar, Zipeng Fu, Deepak Pathak, and Jitendra Malik. RMA : Rapid motor adaptation for legged robots. In Proc.\ Robotics: Science and Systems ( RSS ) , 2021

    2021

  22. [30]

    Algebraic Riccati Equations

    P Lancaster. Algebraic Riccati Equations. Oxford Science Publications/The Clarendon Press, Oxford University Press, 1995

    1995

  23. [31]

    A tutorial on energy-based learning

    Yann LeCun, Sumit Chopra, Raia Hadsell, M Ranzato, and Fujie Huang. A tutorial on energy-based learning. Predicting structured data, 1 0 (0), 2006

    2006

  24. [32]

    Learning quadrupedal locomotion over challenging terrain

    Joonho Lee, Jemin Hwangbo, Lorenz Wellhausen, Vladlen Koltun, and Marco Hutter. Learning quadrupedal locomotion over challenging terrain. Science robotics, 5 0 (47): 0 eabc5986, 2020 a

    2020

  25. [33]

    Making sense of vision and touch: Learning multimodal representations for contact-rich tasks

    Michelle A Lee, Yuke Zhu, Peter Zachares, Matthew Tan, Krishnan Srinivasan, Silvio Savarese, Li Fei-Fei, Animesh Garg, and Jeannette Bohg. Making sense of vision and touch: Learning multimodal representations for contact-rich tasks. IEEE Transactions on Robotics, 36 0 (3): 0 5...

    2020

  26. [34]

    Goal-conditioned reinforcement learning: Problems and solutions

    Minghuan Liu, Menghui Zhu, and Weinan Zhang. Goal-conditioned reinforcement learning: Problems and solutions. In Proc.\ Intl Joint Conf.\ on Artificial Intelligence (IJCAI), pages 5502--5511, 2022

    2022

  27. [35]

    Dynamic tube mpc for nonlinear systems

    Brett T Lopez, Jean-Jacques E Slotine, and Jonathan P How. Dynamic tube mpc for nonlinear systems. In 2019 American Control Conference (ACC), pages 1655--1662. IEEE, 2019

    2019

  28. [36]

    Stable pushing: Mechanics, controllability, and planning

    Kevin M Lynch and Matthew T Mason. Stable pushing: Mechanics, controllability, and planning. International Journal of Robotics Research (IJRR), 15 0 (6): 0 533--556, 1996

    1996

  29. [37]

    Mechanics and planning of manipulator pushing operations

    Matthew T Mason. Mechanics and planning of manipulator pushing operations. International Journal of Robotics Research (IJRR), 5 0 (3): 0 53--71, 1986

    1986

  30. [38]

    Progress in nonprehensile manipulation

    Matthew T Mason. Progress in nonprehensile manipulation. International Journal of Robotics Research (IJRR), 18 0 (11): 0 1129--1141, 1999

    1999

  31. [39]

    Active domain randomization

    Bhairav Mehta, Manfred Diaz, Florian Golemo, Christopher J Pal, and Liam Paull. Active domain randomization. In Conference on Robot Learning, pages 1162--1176. PMLR, 2020

    2020

  32. [40]

    Model predictive control: past, present and future

    Manfred Morari and Jay H Lee. Model predictive control: past, present and future. Computers & chemical engineering, 23 0 (4-5): 0 667--682, 1999

    1999

  33. [41]

    Domain randomization for simulation-based policy optimization with transferability assessment

    Fabio Muratore, Felix Treede, Michael Gienger, and Jan Peters. Domain randomization for simulation-based policy optimization with transferability assessment. In Conference on Robot Learning, pages 700--713. PMLR, 2018

    2018

  34. [42]

    Neural posterior domain randomization

    Fabio Muratore, Theo Gruner, Florian Wiese, Boris Belousov, Michael Gienger, and Jan Peters. Neural posterior domain randomization. In Conference on Robot Learning, pages 1532--1542. PMLR, 2022

    2022

  35. [43]

    Tensor networks for complex quantum systems

    Román Orús. Tensor networks for complex quantum systems. Nature Reviews Physics, 1 0 (9): 0 538--550, 2019. ISSN 2522-5820

    2019

  36. [44]

    TT -cross approximation for multidimensional arrays

    Ivan Oseledets and Eugene Tyrtyshnikov. TT -cross approximation for multidimensional arrays. Linear Algebra and its Applications, 432 0 (1): 0 70--88, 2010

    2010

  37. [45]

    Tensor-train decomposition

    Ivan V Oseledets. Tensor-train decomposition. SIAM Journal on Scientific Computing, 33 0 (5): 0 2295--2317, 2011

    2011

  38. [46]

    Value iteration and policy iteration algorithms for M arkov decision problem

    Elena Pashenkova, Irina Rish, and Rina Dechter. Value iteration and policy iteration algorithms for M arkov decision problem. In AAAI’96: Workshop on Structural Issues in Planning and Temporal Reasoning, volume 39. Citeseer, 1996

    1996

  39. [47]

    Sim-to-real transfer of robotic control with dynamics randomization

    Xue Bin Peng, Marcin Andrychowicz, Wojciech Zaremba, and Pieter Abbeel. Sim-to-real transfer of robotic control with dynamics randomization. In Proc. IEEE Intl Conf. on Robotics and Automation ( ICRA ) , pages 3803--3810, 2018

    2018

  40. [48]

    Accelerating reinforcement learning with learned skill priors

    Karl Pertsch, Youngwoon Lee, and Joseph Lim. Accelerating reinforcement learning with learned skill priors. In Conference on robot learning, pages 188--204. PMLR, 2021

    2021

  41. [49]

    Learning-based control for pushing with a non-holonomic mobile robot

    Susan Potters. Learning-based control for pushing with a non-holonomic mobile robot. Master's thesis, Delft University of Technology, Delft, 2022

    2022

  42. [50]

    Approximate Dynamic Programming: Solving the curses of dimensionality, volume 703

    Warren B Powell. Approximate Dynamic Programming: Solving the curses of dimensionality, volume 703. John Wiley & Sons, 2007

    2007

  43. [51]

    In-hand object rotation via rapid motor adaptation

    Haozhi Qi, Ashish Kumar, Roberto Calandra, Yi Ma, and Jitendra Malik. In-hand object rotation via rapid motor adaptation. In Conference on Robot Learning, pages 1722--1732. PMLR, 2023

    2023

  44. [52]

    Stable-baselines3: Reliable reinforcement learning implementations

    Antonin Raffin, Ashley Hill, Adam Gleave, Anssi Kanervisto, Maximilian Ernestus, and Noah Dormann. Stable-baselines3: Reliable reinforcement learning implementations. Journal of Machine Learning Research, 22 0 (268): 0 1--8, 2021

    2021

  45. [53]

    Bayessim: adaptive domain randomization via probabilistic inference for robotics simulators

    Fabio Ramos, Rafael Carvalhaes Possas, and Dieter Fox. Bayessim: adaptive domain randomization via probabilistic inference for robotics simulators. In Proc.\ Robotics: Science and Systems ( RSS ) , 2019

    2019

  46. [54]

    Savostyanov and Ivan V

    Dmitry V. Savostyanov and Ivan V. Oseledets. Fast adaptive interpolation of multi-dimensional arrays in tensor train format. The 2011 International Workshop on Multidimensional (nD) Systems, pages 1--8, 2011

    2011

  47. [55]

    Is imitation learning the route to humanoid robots? Trends in cognitive sciences, 3 0 (6): 0 233--242, 1999

    Stefan Schaal. Is imitation learning the route to humanoid robots? Trends in cognitive sciences, 3 0 (6): 0 233--242, 1999

    1999

  48. [56]

    Tensor train for global optimization problems in robotics

    Suhan Shetty, Teguh Lembono, Tobias L\"ow, and Sylvain Calinon. Tensor train for global optimization problems in robotics. International Journal of Robotics Research (IJRR), 43 0 (6): 0 811--839, 2024 a . doi:10.1177/02783649231217527

    2024 doi

  49. [57]

    Generalized policy iteration using tensor approximation for hybrid control

    Suhan Shetty, Teng Xue, and Sylvain Calinon. Generalized policy iteration using tensor approximation for hybrid control. In Proc.\ Intl Conf.\ on Learning Representations ( ICLR ) , 2024 b

    2024

  50. [58]

    Robot Learning using Tensor Networks

    Suhan Narayana Shetty. Robot Learning using Tensor Networks. PhD thesis, EPFL, Lausanne, 2024

    2024

  51. [59]

    Sliding mode control and observation, volume 10

    Yuri Shtessel, Christopher Edwards, Leonid Fridman, and Arie Levant. Sliding mode control and observation, volume 10. Springer, 2014

    2014

  52. [60]

    Ttopt: A maximum volume quantized tensor train-based optimization and its application to reinforcement learning

    Konstantin Sozykin, Andrei Chertkov, Roman Schutski, Anh-Huy Phan, Andrzej S Cichocki, and Ivan Oseledets. Ttopt: A maximum volume quantized tensor train-based optimization and its application to reinforcement learning. Advances in neural information processing systems, 35: 0 ...

    2022

  53. [61]

    Supervised learning with tensor networks

    Edwin Stoudenmire and David J Schwab. Supervised learning with tensor networks. Advances in neural information processing systems, 29, 2016

    2016

  54. [62]

    Continuous tensor train-based dynamic programming for high-dimensional zero-sum differential games

    Ezra Tal, Alex Gorodetsky, and Sertac Karaman. Continuous tensor train-based dynamic programming for high-dimensional zero-sum differential games. In 2018 Annual American Control Conference (ACC), pages 6086--6093. IEEE, 2018

    2018

  55. [63]

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

    Josh Tobin, Rachel Fong, Alex Ray, Jonas Schneider, Wojciech Zaremba, and Pieter Abbeel. Domain randomization for transferring deep neural networks from simulation to the real world. In Proc. IEEE/RSJ Intl Conf. on Intelligent Robots and Systems ( IROS ) , pages 23--30, 2017

    2017

  56. [64]

    Behavioral cloning from observation

    Faraz Torabi, Garrett Warnell, and Peter Stone. Behavioral cloning from observation. In Proc.\ Intl Joint Conf.\ on Artificial Intelligence (IJCAI), pages 4950--4957, 2018

    2018

  57. [65]

    Describing physics for physical reasoning: Force-based sequential manipulation planning

    Marc Toussaint, Jung-Su Ha, and Danny Driess. Describing physics for physical reasoning: Force-based sequential manipulation planning. IEEE Robotics and Automation Letters ( RA-L ) , 5 0 (4): 0 6209--6216, 2020

    2020

  58. [66]

    Sequence-of-constraints mpc: Reactive timing-optimal control of sequential manipulation

    Marc Toussaint, Jason Harris, Jung-Su Ha, Danny Driess, and Wolfgang H \"o nig. Sequence-of-constraints mpc: Reactive timing-optimal control of sequential manipulation. In Proc. IEEE/RSJ Intl Conf. on Intelligent Robots and Systems ( IROS ) , pages 13753--13760, 2022

    2022

  59. [67]

    tntorch: Tensor network learning with pytorch

    Mikhail Usvyatsov, Rafael Ballester-Ripoll, and Konrad Schindler. tntorch: Tensor network learning with pytorch. Journal of Machine Learning Research, 23 0 (208): 0 1--6, 2022

    2022

  60. [68]

    Adaptive control for pivoting with visual and tactile feedback

    Francisco Vi \ n a Barrientos, Yiannis Karayiannidis, Christian Smith, and Danica Kragic. Adaptive control for pivoting with visual and tactile feedback. In Proc. IEEE Intl Conf. on Robotics and Automation ( ICRA ) , 2016

    2016

  61. [69]

    Diffusion policies as an expressive policy class for offline reinforcement learning

    Zhendong Wang, Jonathan J Hunt, and Mingyuan Zhou. Diffusion policies as an expressive policy class for offline reinforcement learning. In Proc.\ Intl Conf.\ on Learning Representations ( ICLR ) , 2022

    2022

  62. [70]

    Approximate dynamic programming for real-time control and neural modeling

    Paul Werbos. Approximate dynamic programming for real-time control and neural modeling. Handbook of intelligent control, 1992

    1992

  63. [71]

    Demonstration-guided optimal control for long-term non-prehensile planar manipulation

    Teng Xue, Hakan Girgin, Teguh Santoso Lembono, and Sylvain Calinon. Demonstration-guided optimal control for long-term non-prehensile planar manipulation. In Proc. IEEE Intl Conf. on Robotics and Automation ( ICRA ) , pages 4999--5005, 2023

    2023

  64. [72]

    Robust manipulation primitive learning via domain contraction

    Teng Xue, Amirreza Razmjoo, Suhan Shetty, and Sylvain Calinon. Robust manipulation primitive learning via domain contraction. In Proc.\ Conference on Robot Learning ( CoRL ) , 2024 a

    2024

  65. [73]

    Logic-Skill Programming: An Optimization-based Approach to Sequential Skill Planning

    Teng Xue, Amirreza Razmjoo, Suhan Shetty, and Sylvain Calinon. Logic-Skill Programming: An Optimization-based Approach to Sequential Skill Planning . In Proc.\ Robotics: Science and Systems ( RSS ) , 2024 b

    2024

  66. [74]

    Diffusion models: A comprehensive survey of methods and applications

    Ling Yang, Zhilong Zhang, Yang Song, Shenda Hong, Runsheng Xu, Yue Zhao, Wentao Zhang, Bin Cui, and Ming-Hsuan Yang. Diffusion models: A comprehensive survey of methods and applications. ACM Computing Surveys, 56 0 (4): 0 1--39, 2023

    2023

  67. [75]

    Preparing for the unknown: Learning a universal policy with online system identification

    Wenhao Yu, Jie Tan, C Karen Liu, and Greg Turk. Preparing for the unknown: Learning a universal policy with online system identification. In Proc.\ Robotics: Science and Systems ( RSS ) , 2017

    2017

  68. [76]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution isbn journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence...

  69. [77]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  70. [78]

    , " * write output.state after.block = add.period write newline

    ENTRY address archive author booktitle chapter doi edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type url volume year label INTEGERS output.state before.all mid.sentence after.sentence aft...

  71. [79]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.