Subspace-wise Hybrid RL for Articulated Object Manipulation

Bum-Jae You; Keunwoo Jang; Sol Choi; Yisoo Lee; Yujin Kim

REVIEW 3 major objections 4 minor 26 references

Subspace-wise Hybrid RL for Articulated Object Manipulation

T0 review · 3 major / 4 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 Decomposing articulated-object manipulation into kinematic, geometric, and redundant subspaces, with two learned policies, beats whole-task RL and manual control.

desk verdict The core decomposition idea is real and worth engaging, but the headline result in Table 1 is not reproducible from the paper's own numbers, so it needs a serious referee and a major revision, not a desk reject. read the letter →

arxiv 2412.08522 v1 pith:RYTVNMGC submitted 2024-12-11 cs.RO

Yujin Kim , Sol Choi , Bum-Jae You , Keunwoo Jang , Yisoo Lee This is my paper · ORCID

classification cs.RO

keywords articulatedobjectmanipulationsubspace-wisereinforcementlearninghybridforce/motioncontrolredundantsubspacetaskspacedecompositionadaptiveforcesim-to-real

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

SwRL claims that a robot can manipulate articulated objects more reliably by refusing to learn the whole task at once: it splits the end-effector task space, defined in a frame attached to the object's joint, into a kinematic subspace controlled by force, a geometric subspace controlled by planned motion, and a redundant subspace whose free motion is learned by a second policy. The kinematic policy learns only how much force to apply so the object keeps moving at a desired speed despite unknown friction, damping, or springs; the redundant policy learns how to use extra degrees of freedom to avoid collisions, singularities, and joint limits. The two learned outputs feed a classical hybrid force/motion controller through a selection matrix. The authors report that this decomposition converges faster during training and produces larger articulations than manual control, behavior cloning, and a vanilla RL policy on handwheel valves, lever valves, doors, and drawers, and that a policy trained in simulation turns an unseen real valve. If true, the insight is that the bottleneck in contact-rich manipulation is not the whole control law but the small force-redundancy decision problem wrapped around it.

What carries the argument

The load-bearing mechanism is the object-oriented frame $\{O\}$ placed at the articulated joint, whose $z$-axis is the joint's motion axis, together with the partition $T=S_K\cup S_G\cup S_R$ and the selection matrix $S$ that realizes it inside a hybrid force/motion controller. $S$ routes motion control to the geometric and redundant subspaces and force control to the kinematic subspace, so the $S_K$-policy and $S_R$-policy can be trained separately and executed in parallel at 100 Hz with interpolation into a 1 kHz controller. The $S_R$-policy's reward, which rewards episode length plus penalties for oscillation and collision force, is what turns an otherwise unconstrained subspace into a learned mechanism for avoiding joint limits and singularities; the $S_K$-policy's discretized force increments turn the unknown object dynamics into a one-dimensional adaptive regulation problem. A recurrent network over ten steps of history supplies the Markov state for both policies.

What would settle it

Run SwRL on a handwheel valve whose rotation axis is tilted relative to the assumed object frame, so the true kinematic motion has components in both the kinematic and geometric subspaces; if the robot fails to articulate it while a single-policy RL baseline still makes progress, the pre-fixed decomposition rather than the learning is the limiting factor.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the full six-dimensional task space of an articulated-object manipulation is better treated as three subspaces with different control modalities: kinematic subspace $S_K$, which must be force-controlled and whose force magnitude is the only thing that needs learning; geometric subspace $S_G$, which can be planned from geometry and stays under motion control; and redundant subspace $S_R$, which previous methods froze as geometric constraints but which SwRL actively learns to move in, with the simple objective of keeping the task viable as long as possible. For the handwheel valve the assignment is $S_K=\{x,y\}$, $S_G=\{z,\alpha\}$, $S_R=\{\gamma,\beta\}$. The $S_K$-policy outputs discrete increments to the desired force magnitude, selected from $\{0.1,0,-0.1,1\}$, and is rewarded for holding the object at a target angular velocity; the $S_R$-policy outputs accelerations in the redundant subspace and is rewarded for long episodes, low oscillation, and low collision force. A selection matrix $S$ derived from the decomposition feeds motion commands for $S_G\cup S_R$ and force commands for $S_K$ into the hybrid force/motion control law, so the two learned policies act in parallel without interference. The paper reports faster convergence and better final articulation than manual, behavior-cloning, and single-policy RL baselines across four objects, and a real valve rotation with an unmeasured friction profile.

Load-bearing premise

The decomposition of the task space into kinematic, geometric, and redundant subspaces is fixed in advance from prior knowledge of the object and grasp, and the real object's joint axis and geometry must match that assignment for the force/motion decoupling in the controller to be valid.

Editorial extensions

If this is right

On handwheel valves, lever valves, doors, and drawers, SwRL reaches larger average articulation than manual, behavior-cloning, reinforcement-learning, and ablation baselines, with relative improvements over manual ranging from 17.6% on the lever valve to 67.5% on the drawer.
The learned kinematic force profile lets the robot overcome unknown static friction and recover after a stall, as shown by the real valve where force spikes to roughly 50 N to break static friction and then settles to a steady rotating force.
Using the redundant subspace avoids singularities and collisions that stop manual methods, and keeps the robot's manipulability index higher throughout the motion.
SwRL converges faster during training: the kinematic policy's return rises more quickly than vanilla RL, which often stays stationary and fails to articulate the object while inflating its redundant-space reward through prolonged episodes.
Compared with an offline constrained sampling planner, SwRL completes tasks in real time and opens the door further, although the planner achieves a slightly larger valve rotation after hundreds of seconds of planning.

Reading between the lines

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

The same three-subspace decomposition could apply to other contact-rich tasks with a known constraint manifold, such as peg insertion, screwing, or crank turning, replacing the hand-coded geometric subspace with a geometric prior.
The paper's own stated limitation that the decomposition is pre-determined suggests a testable next step: learn the $S_K/S_G/S_R$ assignment automatically from object category or perception, which would remove the need for prior knowledge of the object and grasp.
The redundant-policy reward of maximizing episode length is a proxy for 'do not lose the task'; richer shaping for energy efficiency or obstacle clearance could improve behavior further, but would require the designer to specify what to optimize.
The vanilla RL baseline's inflated redundant-space reward shows that judging a single policy by summed scalar reward can hide a failure in force control, suggesting that decomposition makes reward design more interpretable and may matter more than raw sample counts.

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.

Desk Editor's Note

read the letter

You should know two things about this one. First, the core idea is genuinely useful: splitting the task space into kinematic, geometric, and redundant subspaces, then learning separate RL policies for force magnitude and redundant motion inside a hybrid force/motion controller, is a real combination I haven't seen in the prior art. The qualitative story holds up—decomposing objectives should help learning, and the redundant subspace is usually ignored. Second, the main quantitative evidence has a reproducibility problem. The stress-test is right: applying the paper's own RMP formula to the averages in Table 1 gives numbers that don't match the table for three of the four objects. For example, handwheel valve: (258.6-202.6)/202.6 = 27.6%, not the reported 31.5%. Lever handle and door are off in the same way. Only the drawer row matches. This could be an aggregation artifact—RMP computed per case and then averaged rather than from the averaged positions—but the paper doesn't say that, and without that explanation the headline claim "SwRL outperforms the baseline methods in all cases" is not reproducible from the paper's own data. That's a load-bearing issue, not a typo.

What's good beyond the idea: the real-world valve experiment is a welcome sanity check, and the authors honestly state the main limitation up front—the subspace decomposition must be pre-determined per object type. That's a real constraint, but they own it. The comparison with CBiRRT is also useful, even if it shows CBiRRT does better on the valve task; the point about planning time is fair.

What's missing: error bars, standard deviations, and significance tests. With only 10 cases for door and drawer, and no variance, the quantitative claims are thin. The paper would be stronger if the authors reported per-case statistics, clarified the RMP aggregation, and released code or data. None of this is fatal to the idea, but it matters for the evidence.

Who should read it: people working on articulated object manipulation or combined force/motion control with RL. The method is plausible and the framework is clearly described. I'd send it to peer review with a request for major revision on the experimental reporting. It deserves serious referee time, not a desk reject.

Referee Report

3 major / 4 minor

Summary. The paper proposes SwRL, a subspace-wise hybrid reinforcement learning framework for articulated object manipulation. The task space is decomposed in an object-oriented frame into a kinematic subspace (SK), a geometric subspace (SG), and a redundant subspace (SR). A SK-policy learns force magnitude commands in the kinematic subspace, while an SR-policy learns acceleration commands in the redundant subspace; the two are integrated through a hybrid force/motion controller. The method is evaluated in MuJoCo simulations on a handwheel valve, lever handle valve, door, and drawer, and in a real-world valve-turning experiment with a Franka Research 3 arm, with comparisons against manual control, behavior cloning, vanilla RL, and the CBiRRT planning method. The central claim is that subspace-wise decomposition improves learning efficiency and manipulation performance and that utilizing redundancy enhances dexterity.

Significance. If the quantitative results were internally consistent, the paper would make a useful contribution by connecting task-space decomposition, hybrid force/motion control, and RL in a way that is both interpretable and deployable on a real robot. The framework's explicit separation of kinematic, geometric, and redundant subspaces is a sensible response to the sample-efficiency problems of end-to-end RL, and the real-world validation on an unseen valve is a concrete strength. The authors also state clearly that the subspace decomposition must be pre-determined, which is an honest limitation. However, the quantitative evidence for the main claim is currently not reproducible: several entries in Table 1 cannot be obtained from the paper's own RMP formula, and no uncertainty or significance measures accompany the reported averages. The comparison with CBiRRT also contradicts the blanket claim that SwRL outperforms all baselines. For these reasons, the paper's significance is conditional on fixing the evidence base.

major comments (3)

[§5.3, Table 1] The Relative Articulated Percentage values in Table 1 do not match the formula RMP = (θ_method − θ_manual)/θ_manual × 100 when applied to the reported average articulation positions. For the handwheel valve, (258.6 − 202.6)/202.6 × 100 = 27.6%, not 31.5%. For the lever handle valve, (173.3 − 159.8)/159.8 × 100 = 8.4%, not 17.6%. For the door, (41.3 − 36.7)/36.7 × 100 = 12.5%, not 26.3%. Only the drawer row (67.5%) is consistent. Several sub-columns are also inconsistent: e.g., for the handwheel valve, SwRL-SR should be 7.4% instead of 15.4%, and for the door, SwRL-SK should be −12.5% instead of −1.7%. The paper does not state whether RMP is computed per case and then averaged, or computed from the displayed averages, or how clipping to [−100, +100] is applied before aggregation. Since the headline claim that SwRL outperforms all baselines rests on these percentages, the metric must be made reproducible and the table corrected or re-derived.
[§5.4, Table 1] The quantitative comparison lacks statistical support. Table 1 reports only average articulation positions over 120 cases for the two valves and 10 cases for the door and drawer, with no standard deviations, confidence intervals, or significance tests. The claim that SwRL 'outperformed the baseline methods in all cases' cannot be assessed from a single mean per condition, especially when the manual method's door average is 36.7° and SwRL's is 41.3° with unknown spread. The authors should report per-case results, error bars, and at least pairwise significance tests, or moderate the claim accordingly.
[§5.6 and §6] The statement in Section 6 that the method 'outperforms methods that learn or plan the full task space' is contradicted by the authors' own comparison in §5.6, where CBiRRT achieves 297.9° on the valve-turning task versus SwRL's 272.2°. If CBiRRT is considered a baseline, then SwRL does not outperform all baselines; if it is not, the conclusion should be qualified to the four baselines in Table 1. The paper should reconcile this discrepancy or explicitly characterize SwRL as comparable to, but not superior to, CBiRRT on the valve task.

minor comments (4)

[§4.2.2] The notation '△F ∈ I4' is undefined; the text lists four discrete values, so the symbol I4 should be clarified or replaced with a concrete discrete set.
[§5.2] The behavior cloning baseline description says 'The agent receives inputs as described in Section 5.2', but the state representation is defined in Section 5.1 and Section 4.2.1; the cross-reference should be corrected.
[§5.6] The text says 'an RPM of −12.8%' in the door-opening task; this appears to be a typo for 'RMP'.
[Table 1] The table uses inconsistent formatting, such as '0 .0%' and '31.78◦'; these should be cleaned up to '0.0%' and '31.8°'.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the central claim is grounded in external baselines and real-world transfer; a minor non-load-bearing self-citation and an acknowledged decomposition assumption do not make the derivation circular.

full rationale

The paper's central claim, that subspace-wise hybrid RL improves articulated object manipulation, is supported by comparisons against external baselines (Manual, Behavioral Cloning, vanilla RL, and CBiRRT) and by real-world valve experiments, so it does not reduce to its own inputs by construction. The only self-citation, reference [9] by co-author Jang, appears in related-work examples of pose-constrained manipulation; it is not invoked as a uniqueness theorem and the same point is independently supported by references [7] and [17], so it is not load-bearing. The task-space decomposition into SK, SG, and SR in Section 4.1 is a modeling choice rather than a derived prediction, and the SK-policy reward in Section 4.2.3 encodes the desired joint velocity as a control objective, which is reward design rather than circularity. The conclusion's stated limitation that the decomposition must be pre-determined is an acknowledged input assumption, not a circular step. Separately, Section 5.3's RMP formula is not reproducible from three rows of Table 1, which is a reproducibility and correctness concern outside the scope of circularity and does not raise the circularity score here.

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

The central claim rests on manually chosen control targets, reward weights, and a hand-specified subspace decomposition. These are not derived from first principles, and the method's success depends on them. No new physical entities are postulated; the 'object-oriented frame' is a coordinate convention, not an invented entity.

free parameters (5)

Desired joint velocity targets per object = valves: 0.7-0.8 rad/s, doors: 0.1-0.15 rad/s, drawers: 0.4-0.5 m/s
Hand-set desired velocities used in the SK reward; they define what the force policy must achieve and vary by object class (Section 4.2.3).
Reward weights k1, k2 = k1=1, k2=0.1
Hand-chosen weights in RSR to penalize oscillation and collisions (Section 4.2.3).
Delta force action set = {0.1, 0, -0.1, 1}
Discretized force increments selected by the SK-policy; chosen for training stability (Section 4.2.2).
Subspace assignment per object type = e.g., handwheel valve: SK={x,y}, SG={z,alpha}, SR={gamma,beta}
The a priori partition of the 6D task space is chosen by hand based on the object's joint type and geometry (Section 4.1).
Offline/online data mixing ratio = 1:1
The ratio of pre-collected manual data to online exploration data used for the valve task (Section 4.4).

assumptions (5)

standard math Hybrid force/motion control (Eq. 1) is valid, i.e., the task space can be split into orthogonal position- and force-controlled parts.
Adopted from Raibert and Craig [15]; the selection matrix S enforces independence (Sections 3.2, 4.3).
domain assumption The object is a single-joint articulated body (one revolute or prismatic joint) with no other displacement.
Stated as a characterizing assumption in Section 3; limits the method's scope.
domain assumption Object joint velocity can be estimated from end-effector position tracking without slip or deformation.
Section 4.2.1: 'the exact velocity of the object during manipulation cannot be measured directly. Instead, omega and v were estimated by tracking the position of the robot's end-effector.' The SK reward and real-world deployment depend on this.
domain assumption Simulation-trained policies transfer to the real robot without additional adaptation or domain randomization.
Section 5.5 applies the simulated policy directly to a real valve, with only qualitative success reported.
ad hoc to paper The hand-specified rewards (RSK, RSR) capture the true task objectives (desired joint velocity, avoid oscillation, avoid collisions).
Reward design is arbitrary and not derived from first principles; results depend on these choices (Section 4.2.3).

how reviews work

0 comments

Cite this review

Pith. "Pith review of Subspace-wise Hybrid RL for Articulated Object Manipulation." pith.science (2026). https://pith.science/paper/RYTVNMGC

@misc{pith2026241208522,
  author       = {Pith},
  title        = {Pith review of: Subspace-wise Hybrid RL for Articulated Object Manipulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RYTVNMGC}},
  note         = {Machine review of arXiv:2412.08522}
}

read the original abstract

Articulated object manipulation is a challenging task, requiring constrained motion and adaptive control to handle the unknown dynamics of the manipulated objects. While reinforcement learning (RL) has been widely employed to tackle various scenarios and types of articulated objects, the complexity of these tasks, stemming from multiple intertwined objectives makes learning a control policy in the full task space highly difficult. To address this issue, we propose a Subspace-wise hybrid RL (SwRL) framework that learns policies for each divided task space, or subspace, based on independent objectives. This approach enables adaptive force modulation to accommodate the unknown dynamics of objects. Additionally, it effectively leverages the previously underlooked redundant subspace, thereby maximizing the robot's dexterity. Our method enhances both learning efficiency and task execution performance, as validated through simulations and real-world experiments. Supplementary video is available at https://youtu.be/PkNxv0P8Atk

Figures

Figures reproduced from arXiv: 2412.08522 by the authors.

**Figure 2.** Illustration of object-oriented frame of various objects. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Overview of the Subspace-wise hybrid RL (SwRL) framework for articulated object manipulation. Task [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (6 more)

**Figure 4.** Figure 4: Illustration of geometric constraints for a [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 6.** Figure 6: Comparison of SwRL and manual method across four tasks. The blue boxes and labels indicate the terminal [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 7.** Figure 7: The episode return during the training process for each task is illustrated. The [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

**Figure 8.** Figure 8: Comparison of the manipulability measure between the proposed method and the manual approach as a [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]

**Figure 9.** Figure 9: Real-world experiment settings on three different valve configurations. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]

**Figure 10.** Figure 10: Valve turning motion and corresponding force profile with an unseen real-world valve. [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 19 canonical work pages

[1]

Maniskill: Learning-from-demonstrations benchmark for generalizable manipulation skills

Tongzhou Mu, Zhan Ling, Fanbo Xiang, Derek Yang, Xuanlin Li, Stone Tao, Zhiao Huang, Zhiwei Jia, and Hao Su. Maniskill: Learning-from-demonstrations benchmark for generalizable manipulation skills. CoRR, abs/2107.14483, 2021b. URL https://arxiv. org/abs/2107, 14483, 2021

arXiv 2021
[2]

Universal manipulation policy network for articulated objects

Zhenjia Xu, Zhanpeng He, and Shuran Song. Universal manipulation policy network for articulated objects. IEEE Robotics and Automation Letters, 7(2):2447–2454, 2022

work page 2022
[3]

Flowbot3d: Learning 3d articulation flow to manipulate articulated objects

Ben Eisner, Harry Zhang, and David Held. Flowbot3d: Learning 3d articulation flow to manipulate articulated objects. arXiv preprint arXiv:2205.04382, 2022

arXiv 2022
[4]

Unidoormanip: Learning universal door manipulation policy over large-scale and diverse door manipulation environments

Yu Li, Xiaojie Zhang, Ruihai Wu, Zilong Zhang, Yiran Geng, Hao Dong, and Zhaofeng He. Unidoormanip: Learning universal door manipulation policy over large-scale and diverse door manipulation environments. arXiv preprint arXiv:2403.02604, 2024

arXiv 2024
[5]

Learning force control for contact-rich manipulation tasks with rigid position-controlled robots

Cristian Camilo Beltran-Hernandez, Damien Petit, Ixchel Georgina Ramirez-Alpizar, Takayuki Nishi, Shinichi Kikuchi, Takamitsu Matsubara, and Kensuke Harada. Learning force control for contact-rich manipulation tasks with rigid position-controlled robots. IEEE Robotics and Automation Letters , 5(4):5709–5716, 2020

work page 2020
[6]

task frame formalism

Herman Bruyninckx and Joris De Schutter. Specification of force-controlled actions in the" task frame formalism"- a synthesis. IEEE Transactions on Robotics and Automation , 12(4):581–589, 1996

work page 1996
[7]

Task space regions: A framework for pose-constrained manipulation planning

Dmitry Berenson, Siddhartha Srinivasa, and James Kuffner. Task space regions: A framework for pose-constrained manipulation planning. The International Journal of Robotics Research , 30(12):1435–1460, 2011

work page 2011
[8]

Path planning under kinematic constraints by rapidly exploring manifolds

Léonard Jaillet and Josep M Porta. Path planning under kinematic constraints by rapidly exploring manifolds. IEEE Transactions on Robotics, 29(1):105–117, 2012

work page 2012

Show all 26 references

[9]

Motion planning of mobile manipulator for navigation including door traversal

Keunwoo Jang, Sanghyun Kim, and Jaeheung Park. Motion planning of mobile manipulator for navigation including door traversal. IEEE Robotics and Automation Letters , 2023. 13 A PREPRINT SUBMITTED TO ROBOTICS AND AUTONOMOUS SYSTEMS - D ECEMBER 12, 2024

2023
[10]

Rrt-connect: An efficient approach to single-query path planning

James J Kuffner and Steven M LaValle. Rrt-connect: An efficient approach to single-query path planning. In Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No. 00CH37065) , volume 2, pages 995–10...

2000
[11]

Adaptive tracking control for robots with unknown kinematic and dynamic properties

Chien-Chern Cheah, Chao Liu, and Jean-Jacques E Slotine. Adaptive tracking control for robots with unknown kinematic and dynamic properties. The International Journal of Robotics Research , 25(3):283–296, 2006

2006
[12]

Articulated object interaction in unknown scenes with whole-body mobile manipulation

Mayank Mittal, David Hoeller, Farbod Farshidian, Marco Hutter, and Animesh Garg. Articulated object interaction in unknown scenes with whole-body mobile manipulation. In 2022 IEEE/RSJ international conference on intelligent robots and systems (IROS), pages 1647–1654. IEEE, 2022

2022
[13]

Pulling open doors and drawers: Coordinating an omni-directional base and a compliant arm with equilibrium point control

Advait Jain and Charles C Kemp. Pulling open doors and drawers: Coordinating an omni-directional base and a compliant arm with equilibrium point control. In 2010 IEEE International Conference on Robotics and Automation (ICRA), pages 1807–1814. IEEE, 2010

2010
[14]

Sim2real 2: Actively building explicit physics model for precise articulated object manipulation

Liqian Ma, Jiaojiao Meng, Shuntao Liu, Weihang Chen, Jing Xu, and Rui Chen. Sim2real 2: Actively building explicit physics model for precise articulated object manipulation. In 2023 IEEE International Conference on Robotics and Automation (ICRA) , pages 11698–11704. IEEE, 2023

2023
[15]

Hybrid position/force control of manipulators

Marc H Raibert and John J Craig. Hybrid position/force control of manipulators. 1981

1981
[16]

Compliance and force control for computer controlled manipulators

Matthew T Mason. Compliance and force control for computer controlled manipulators. IEEE Transactions on Systems, Man, and Cybernetics , 11(6):418–432, 1981

1981
[17]

An adaptive control approach for opening doors and drawers under uncertainties

Yiannis Karayiannidis, Christian Smith, Francisco Eli Vina Barrientos, Petter Ögren, and Danica Kragic. An adaptive control approach for opening doors and drawers under uncertainties. IEEE Transactions on Robotics, 32(1):161–175, 2016

2016
[18]

Reinforcement learning: An introduction

Richard S Sutton and Andrew G Barto. Reinforcement learning: An introduction . MIT press, 2018

2018
[19]

A unified approach for motion and force control of robot manipulators: The operational space formulation

Oussama Khatib. A unified approach for motion and force control of robot manipulators: The operational space formulation. IEEE Journal on Robotics and Automation , 3(1):43–53, 1987

1987
[20]

Long short-term memory

Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory. Neural Computation, 9(8):1735–1780, 1997

1997
[21]

Efficient online reinforcement learning with offline data

Philip J Ball, Laura Smith, Ilya Kostrikov, and Sergey Levine. Efficient online reinforcement learning with offline data. In International Conference on Machine Learning , pages 1577–1594. PMLR, 2023

2023
[22]

Mujoco: A physics engine for model-based control

Emanuel Todorov, Tom Erez, and Yuval Tassa. Mujoco: A physics engine for model-based control. In 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 5026–5033. IEEE, 2012

2012
[23]

pybind11 – seamless operability between c++11 and python, 2017

Wenzel Jakob, Jason Rhinelander, and Dean Moldovan. pybind11 – seamless operability between c++11 and python, 2017. https://github.com/pybind/pybind11

2017
[24]

Controlling overestimation bias with truncated mixture of continuous distributional quantile critics

Arsenii Kuznetsov, Pavel Shvechikov, Alexander Grishin, and Dmitry Vetrov. Controlling overestimation bias with truncated mixture of continuous distributional quantile critics. In International Conference on Machine Learning, pages 5556–5566. PMLR, 2020

2020
[25]

Manipulability of robotic mechanisms

Tsuneo Yoshikawa. Manipulability of robotic mechanisms. The International Journal of Robotics Research , 4(2):3–9, 1985

1985
[26]

Manipulation planning on constraint manifolds

Dmitry Berenson, Siddhartha S Srinivasa, Dave Ferguson, and James J Kuffner. Manipulation planning on constraint manifolds. In 2009 IEEE international conference on robotics and automation , pages 625–632. IEEE, 2009. 14

2009

Pith tools

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

[1] [1]

Maniskill: Learning-from-demonstrations benchmark for generalizable manipulation skills

Tongzhou Mu, Zhan Ling, Fanbo Xiang, Derek Yang, Xuanlin Li, Stone Tao, Zhiao Huang, Zhiwei Jia, and Hao Su. Maniskill: Learning-from-demonstrations benchmark for generalizable manipulation skills. CoRR, abs/2107.14483, 2021b. URL https://arxiv. org/abs/2107, 14483, 2021

arXiv 2021

[2] [2]

Universal manipulation policy network for articulated objects

Zhenjia Xu, Zhanpeng He, and Shuran Song. Universal manipulation policy network for articulated objects. IEEE Robotics and Automation Letters, 7(2):2447–2454, 2022

work page 2022

[3] [3]

Flowbot3d: Learning 3d articulation flow to manipulate articulated objects

Ben Eisner, Harry Zhang, and David Held. Flowbot3d: Learning 3d articulation flow to manipulate articulated objects. arXiv preprint arXiv:2205.04382, 2022

arXiv 2022

[4] [4]

Unidoormanip: Learning universal door manipulation policy over large-scale and diverse door manipulation environments

Yu Li, Xiaojie Zhang, Ruihai Wu, Zilong Zhang, Yiran Geng, Hao Dong, and Zhaofeng He. Unidoormanip: Learning universal door manipulation policy over large-scale and diverse door manipulation environments. arXiv preprint arXiv:2403.02604, 2024

arXiv 2024

[5] [5]

Learning force control for contact-rich manipulation tasks with rigid position-controlled robots

Cristian Camilo Beltran-Hernandez, Damien Petit, Ixchel Georgina Ramirez-Alpizar, Takayuki Nishi, Shinichi Kikuchi, Takamitsu Matsubara, and Kensuke Harada. Learning force control for contact-rich manipulation tasks with rigid position-controlled robots. IEEE Robotics and Automation Letters , 5(4):5709–5716, 2020

work page 2020

[6] [6]

task frame formalism

Herman Bruyninckx and Joris De Schutter. Specification of force-controlled actions in the" task frame formalism"- a synthesis. IEEE Transactions on Robotics and Automation , 12(4):581–589, 1996

work page 1996

[7] [7]

Task space regions: A framework for pose-constrained manipulation planning

Dmitry Berenson, Siddhartha Srinivasa, and James Kuffner. Task space regions: A framework for pose-constrained manipulation planning. The International Journal of Robotics Research , 30(12):1435–1460, 2011

work page 2011

[8] [8]

Path planning under kinematic constraints by rapidly exploring manifolds

Léonard Jaillet and Josep M Porta. Path planning under kinematic constraints by rapidly exploring manifolds. IEEE Transactions on Robotics, 29(1):105–117, 2012

work page 2012