Leveraging Extrinsic Dexterity for Occluded Grasping on Grasp Constraining Walls

Keita Kobashi; Masayoshi Tomizuka

REVIEW 3 major objections 7 minor 27 references

Leveraging Extrinsic Dexterity for Occluded Grasping on Grasp Constraining Walls

T0 review · 3 major / 7 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper shows that a simulation-trained hierarchy of pivoting, pushing, and grasping skills, coordinated by a DQN and a conditional variational autoencoder, solves occluded grasping on tall grasp-constraining walls, reaching 100%…

desk verdict Solid simulation results and a genuine tall-wall twist, but the real-world experiments sidestep the defining scenario and the high-level policy is less autonomous than claimed. read the letter →

arxiv 2507.14721 v1 pith:SUG3M5T4 submitted 2025-07-19 cs.RO

Keita Kobashi , Masayoshi Tomizuka This is my paper

classification cs.RO

keywords occludedgraspingextrinsicdexterityhierarchicalreinforcementlearningpivot-push-graspconditionalvariationalautoencodersim-to-realtransfergraspconstrainingwallsdomainrandomization

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 paper tackles occluded grasping, where an object's primary graspable faces are blocked by the environment, and specifically the harder case of tall 'grasp constraining' walls where reorienting the object by pivoting still leaves it ungraspable. The authors claim that a hierarchical policy—a high-level DQN that selects among a learned pivoting skill, a hand-coded pushing skill, and a hand-coded grasping skill, with a conditional variational autoencoder predicting contact locations—can combine reorientation and repositioning to make such objects graspable, with no human demonstrations. In simulation the method reaches 100% success on constraining walls while all baselines except a point-cloud variant score 0%; in the real world it transfers zero-shot to six out-of-distribution objects with an average success rate of 90%. The significance is that prior extrinsic-dexterity work assumed grasp-accessible walls, so this extends extrinsic dexterity to environments where a single pivot action is not enough.

What carries the argument

The load-bearing mechanism is the three-part hierarchy that separates discrete task selection from continuous control. A DQN high-level policy observes the object pose and wall length and outputs one of three skill IDs (pivoting, pushing, grasping), with a reward that penalizes selecting a completed subtask or grasping too early. The pivoting skill is a SAC policy in continuous end-effector velocity space trained with domain randomization (Gaussian noise on force and pose); pushing and grasping are hand-coded constant-velocity motions. A conditional variational autoencoder (CVAE), conditioned on the object point cloud and the skill ID, reconstructs contact locations from successful rollouts, and the framework picks the lowest reconstructed point for pivoting and the largest-y point for pushing. This lets the policy re-attempt failed pivots and shift the object along the wall until the grasp is accessible.

What would settle it

Run the trained policy in the real world on a genuinely tall, grasp-constraining wall (e.g., with a side-mounted camera or a transparent wall so vision is not blocked), repeat 30 trials per object, and compare success rates with the short-wall results. If tall-wall success drops well below the 90% short-wall average, or below the 100% simulation rate, the claim that the framework solves grasp constraining walls is falsified; the same test can be done in simulation by increasing wall height beyond the training range.

Watch

Extended reading notes

Core claim

The central claim is that occluded grasping on grasp constraining walls is not solved by pivoting alone, but becomes solvable when the robot can autonomously sequence pivoting, pushing, and grasping, and can choose contact locations from object geometry. Concretely, the paper reports that on tall walls in simulation, the proposed hierarchical framework succeeds in 100% of 30 episodes for boxes and gear objects across wall lengths 0.1–0.4 m, while the SAC baseline, the method without the CVAE, the method without skills, and the Ungraspable baseline all achieve 0% (the Point Cloud baseline achieves 13.3%). In physical experiments, the same simulation-trained policies achieve a 90% average success rate over six unseen objects (boxes, bottles, and a circular object), using an RGB-D camera, AprilTag pose estimation, and admittance control, with no retraining or human demonstrations.

Load-bearing premise

The real-world experiments use a short wall, and the paper assumes that wall height does not change the physical interaction because contact happens on the lower side; if that assumption is wrong, the 90% success rate does not test the tall-wall scenario that the paper's central claim is about.

Editorial extensions

If this is right

On grasp-accessible walls the framework matches the best baselines at 100% success, so adopting the hierarchy does not sacrifice performance where simpler pivoting already works.
On grasp-constraining walls the 0% scores of the SAC pivoting baseline and the emergent extrinsic dexterity baseline imply that a single reorientation action is insufficient; the combination of push and pivot is necessary.
Longer walls increase how often the high-level policy selects the pushing skill, showing the skill selector uses the wall-length observation to adapt its strategy rather than executing a fixed sequence.
The CVAE-based contact selection outperforms the point-cloud baseline on constraining walls (100% vs 13.3%), indicating that generative contact-location inference is a critical component for coordinated multi-skill manipulation.
Zero-shot sim-to-real transfer over six unseen objects, including curved bottles and a circle, indicates that the domain randomization and force-based observations make the learned pivoting skill robust to novel geometry.

Reading between the lines

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

A testable corollary the authors do not draw: because the real-world wall was short, the 90% figure does not by itself validate the tall-wall scenario; running the same policy on a genuinely tall wall (with a camera placed to avoid occlusion) would directly test whether the assumed height-invariance of the contact physics holds.
The framework's reliance on hand-coded push and grasp skills and on fixed heuristics for picking contact points (lowest z for pivot, largest y for push) suggests that the CVAE may be doing less work than the hierarchy's other components; a learned selection rule could reveal whether this hand-crafting is the actual bottleneck.
The authors' future-work suggestion of bins with constraining walls is the natural stress test: in a bin the object can contact multiple walls, and the wall-length observation would need to be replaced by a richer representation, so it is unclear whether the same DQN state space would suffice.

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

Colleague, the short version: this paper has a genuinely useful core — a hierarchical policy that learns to pivot, push, and grasp on tall constraining walls, no demonstrations, zero-shot to unseen objects — and its simulation results are striking: 100% where the strongest baselines get 0–13%. The real-world 90% across six objects is solid evidence of transfer, even if the experiments don't quite test what the title promises.

What's new: prior pivot-then-grasp work assumes short, grasp-accessible walls. This paper tackles the tall-wall case, where pivoting alone leaves the object ungraspable, and learns a high-level DQN that switches among pivoting, pushing, and grasping. The CVAE contact-location estimator conditioned on skill ID is a reasonable addition. Training without human demos and transferring to bottles and a circular object is a real step past [7] and [10].

Soft spots, in rough order of importance. First, the real-world wall is short. Section V-B justifies this with an assertion that the physical interaction happens on the lower side of the wall, but that is not a measurement. On a short wall the object may become graspable after pivoting alone, so the 90% may not exercise the constraining-wall regime at all. The tall-wall claim rests entirely on simulation. Second, the high-level policy's autonomy is overstated. The Eq. (1) penalties are so strong they essentially hard-code the sequence pivot → push → grasp; the DQN is choosing when to repeat a skill, not really choreographing novel sequences. That is a minor overclaim if the fixed order is the right one, but it should be acknowledged. Third, there is no scripted baseline. Given the hand-coded push/grasp and the near-fixed ordering, a sequence of hard-coded triggers would be a cheap and informative comparison. Also, the action rescale factor and CVAE offset vector are mentioned but not reported, making sim-to-real reproduction harder than it needed to be.

The math and citations look fine. The simulation results are internally consistent, and the paper does not fit constants to success rates.

Recommendation: worth a serious referee. The central scenario is under-tested in the real world and the autonomy claim needs softening, but the method and the simulation evidence are real. I would engage with it, ask for the scripted baseline and the tall-wall measurement or a clear justification, then likely accept.

Referee Report

3 major / 7 minor

Summary. The paper proposes a hierarchical reinforcement-learning framework for occluded grasping against tall 'grasp constraining' walls. A DQN high-level policy selects among three low-level skills: a SAC-trained pivoting skill, a hand-coded pushing skill, and a hand-coded grasping skill. A CVAE, conditioned on object point clouds and the skill ID, infers contact locations for pivoting and pushing. The policies are trained entirely in simulation on a box-like object and then deployed zero-shot to six real-world objects. The paper reports 100% success on grasp-constraining walls in simulation (Table I), high success rates on unseen simulated objects across wall lengths (Table II), and a 90% average real-world success rate (Table III), concluding that the framework generalizes and transfers to unseen objects without human demonstrations.

Significance. If the claims as stated were fully supported, the paper would be a useful step toward single-arm occluded grasping on constraining walls: it combines primitive skills with a skill-selection layer, uses a skill-conditioned CVAE for contact locations, reports internal simulation ablations, reports real-world per-object success counts, and does not rely on human demonstrations. These are real strengths. However, two load-bearing issues reduce the significance of the results as currently presented: the real-world validation uses a short wall rather than the tall 'grasp constraining' wall that defines the problem, and the high-level policy's skill ordering is almost entirely fixed by hand-written penalties in Eq. (1), so the claimed 'automatic switching' is not actually demonstrated. The paper is not circular in the derivation sense, since no constants are fit to the reported success rates, but the empirical support for the central tall-wall sim-to-real claim is incomplete.

major comments (3)

[Section V-B (Table III)] The physical experiments use a short wall built from two magnet bases, and the paper justifies this by asserting that 'both on a tall wall and a short wall, the actual physical interaction happens on the lower side of the wall,' so performance is unchanged. This is an unverified equivalence assumption. In Section III, a grasp-constraining wall is defined as tall, i.e., the object remains ungraspable after pivoting; on a short wall the object may become graspable immediately after pivoting, so the real-world trajectories may never enter the constraining regime in which additional pushing is required. Table III therefore supports generalization to unseen objects on a short wall, not the paper's central tall-wall scenario. The authors should either run real experiments with a tall wall (e.g., repositioning the camera or using fiducial pose estimates to avoid occlusion) or provide logged evidence from the short-wall trials that the constraining-wall regime was actually exercised. If neither is possible, the abstract and conclusion should be revised to state that the tall-wall claim is supported only by simulation.
[Section IV-A, Eq. (1)] The reward penalties in Eq. (1) encode a fixed legal action sequence: pivoting is penalized while vpush=0, pushing is penalized once vpush=1, and grasping is penalized unless vpivot*vpush=1. Consequently, in any given state at most one skill (apart from retries of a failed skill) is non-penalized, and the DQN high-level policy is not learning to switch among skills; it executes a hand-authored push-then-pivot-then-grasp script. This conflicts with the abstract and Section I claim that the framework 'automatically switches' between different types of actions. The paper should either retrain or evaluate the high-level policy with the penalties relaxed or removed, or include environments in which the optimal skill order is not fixed, so that learned switching is actually demonstrated. Otherwise the wording should be changed to acknowledge that the ordering is specified by the reward design.
[Section IV-C and Section V] The CVAE's role is to infer contact locations, yet the final contact point is chosen by two hand-specified ranking rules ('lowest z position' for pivoting, 'largest y position' for pushing) plus an unspecified offset vector, and no ablation or sensitivity analysis is reported for these rules. In addition, the paper states that x_goal is computed from the wall length l but never gives the formula, and the sim-to-real action rescale factor is not reported. Since contact-location selection is one of the four stated difficulties and a central contribution, these omissions prevent the reader from distinguishing the contribution of the CVAE from that of the manual selection rules, and they hamper reproduction. Please report the exact definitions and add ablations, such as random selection from CVAE outputs, varying the offset vector, or removing the ranking rules.

minor comments (7)

[General] There are several grammar errors in the text, including 'an grasp accessible environment' and 'on an grasp constraining wall'; these should be corrected.
[Section IV-C] The notation O ∈ R^{n×m} is confusing: the text later refers to a flattened point cloud of length m, and n is used for the number of point clouds and contact locations. Please define n and m explicitly and clarify whether O is a batch of point clouds or a single cloud.
[Section V-A, Eq. (1)] The penalty condition in Eq. (1) is ambiguous because of operator precedence; please parenthesize the disjunctions so that the intended conditions are unambiguous.
[Section V-A, Table II] The dimensions for the gear object, such as '0.062π × 0.026 [m3]' and '0.05452π × 0.026', are unclear; please use explicit radius and height values.
[Section V-B, Table III] The real-world protocol does not state how many attempts were allowed per object or what the failure modes were. Please report the number of trials per object and, where possible, describe the failure cases.
[Section V-A, Fig. 5] The skill-selection frequency plot has no error bars or indication of variance across the 30 episodes; please add this information and clarify whether the frequencies are normalized per episode.
[Section II-A] The related work discusses [8] as addressing grasp-constraining walls, but no experimental comparison to [8] is included in Section V; please either add such a comparison or soften the implicit claim of superiority.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the paper's empirical success claims are validated by simulation and zero-shot real-world experiments, with no fitted parameter renamed as a prediction and no load-bearing self-citation.

full rationale

The paper's central claim is an empirical one: a hierarchical RL framework achieves high success rates on occluded grasping against grasp-constraining walls. The derivation chain is not circular. The high-level DQN, low-level SAC pivoting skill, hand-coded push/grasp skills, and CVAE contact-location model are trained in simulation and then transferred zero-shot to real objects; the real-world success rates in Table III are measured independently of the training objectives, not read off from the reward or from a fitted parameter. The reward shaping in Eq. (1), including the penalties that strongly constrain skill selection, is a legitimate RL design choice rather than a circular step: it does not define the evaluation metric (task success) in terms of the policy's own outputs. Likewise, the CVAE is trained on successful skill rollouts, but its role is contact-location inference, and the final grasping success is not a reconstruction of that training data. The baseline comparisons use external or published implementations (e.g., the provided implementation of [7]), and self-citations to prior work from the same group are related-work references rather than load-bearing premises. One limitation worth flagging is in Section V-B, where the real-world wall is short because a tall wall would cause vision occlusion, and the paper asserts: 'both on a tall wall and a short wall, the actual physical interaction happens on the lower side of the wall. Hence, the performance of the trained RL policy remains unchanged.' This is an unverified equivalence assumption that limits how directly the 90% real-world success rate supports the tall-wall scenario; it is a validity concern, not a circularity concern, because it does not reduce any claim to its own inputs by definition. Overall, the paper is self-contained against external benchmarks and exhibits no significant circularity.

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

The central claim rests on several unverified assumptions about the simulation, the wall, and the transfer procedure. The most important are listed above.

free parameters (5)

High-level reward weights and criteria (r_bonus=0.05, r_penalty=-30, r_done=1.05, pivoting buffer 10 deg) = 0.05, -30, 1.05, 10 deg
Eq. (1) in Sec. IV-A. The -30 penalty makes the skill sequence nearly deterministic, so the high-level policy's behavior is hand-designed rather than learned.
Contact location selection rule for CVAE output = lowest z for pivoting; largest y for pushing
Sec. IV-C: 'We empirically find that the lowest z position is effective for pivoting action, and the largest y position is effective for pushing action.' This is a manual post-processing of the generative model, not a learned selection.
Contact location offset vector = not reported
Sec. V-B: 'When we run the CVAE while executing the whole pipeline, we add an offset vector to the inferred contact location.' The vector is not specified, so it is an unreported tuned value.
Action rescale factor for sim-to-real = not reported
Sec. V-B: 'we uniformly rescale the action from the low-level skills' to compensate for the controller difference. The scale is not given.
Domain randomization noise scales = 0.2 std on fext/xeef/oeef; 0.003 std on point cloud
Sec. IV-B and IV-C. These values are chosen by hand; they are the only randomization used, and there is no physics parameter randomization.

assumptions (5)

domain assumption A short wall interacts with the object identically to a tall wall for the purposes of the low-level physical interaction
Sec. V-B states this explicitly when explaining why the real-world test uses a short wall. If false, the real experiments do not test the grasp-constraining wall scenario.
domain assumption Training on box-like objects with only observation-noise domain randomization and no physics randomization transfers zero-shot to bottles and curved objects
Sec. IV-B and V-B. The sim-to-real success depends on this; curved surfaces are acknowledged to make contact maintenance difficult.
domain assumption The hand-coded pushing and grasping skills are sufficient and robust across object shapes
Sec. IV-B: pushing and grasping are fixed-velocity actions. The framework's success on constrained walls assumes these simple actions achieve the required contact behavior.
domain assumption MuJoCo's rigid-body contact model accurately captures the contact dynamics of pivoting a box against a fixed wall
Sec. V-A uses MuJoCo. The simulation results are the main evidence for the constraining-wall scenario, since the real wall is short.
domain assumption AprilTag pose estimation and color-based point cloud segmentation are accurate enough in deployment
Sec. V-B: the real system depends on AprilTag and color filtering to obtain object pose and point cloud.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Leveraging Extrinsic Dexterity for Occluded Grasping on Grasp Constraining Walls." pith.science (2026). https://pith.science/paper/SUG3M5T4

@misc{pith2026250714721,
  author       = {Pith},
  title        = {Pith review of: Leveraging Extrinsic Dexterity for Occluded Grasping on Grasp Constraining Walls},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SUG3M5T4}},
  note         = {Machine review of arXiv:2507.14721}
}

read the original abstract

This study addresses the problem of occluded grasping, where primary grasp configurations of an object are not available due to occlusion with environment. Simple parallel grippers often struggle with such tasks due to limited dexterity and actuation constraints. Prior works have explored object pose reorientation such as pivoting by utilizing extrinsic contacts between an object and an environment feature like a wall, to make the object graspable. However, such works often assume the presence of a short wall, and this assumption may not always hold in real-world scenarios. If the wall available for interaction is too large or too tall, the robot may still fail to grasp the object even after pivoting, and the robot must combine different types of actions to grasp. To address this, we propose a hierarchical reinforcement learning (RL) framework. We use Q-learning to train a high-level policy that selects the type of action expected to yield the highest reward. The selected low-level skill then samples a specific robot action in continuous space. To guide the robot to an appropriate location for executing the selected action, we adopt a Conditional Variational Autoencoder (CVAE). We condition the CVAE on the object point cloud and the skill ID, enabling it to infer a suitable location based on the object geometry and the selected skill. To promote generalization, we apply domain randomization during the training of low-level skills. The RL policy is trained entirely in simulation with a box-like object and deployed to six objects in real world. We conduct experiments to evaluate our method and demonstrate both its generalizability and robust sim-to-real transfer performance with promising success rates.

Figures

Figures reproduced from arXiv: 2507.14721 by the authors.

**Figure 2.** The overview of our hierarchical reinforcement learning framework for occluded grasping tasks. The high-level policy decides which low-level skill [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** An example for a step of executing the proposed framework. The [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: An example of a successful action sequence of the robot in MuJoCo [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: A mean frequency of skill selection for different lengths of the grasp [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

**Figure 6.** Figure 6: Test Objects for Real-world Experiments. All objects are out [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

**Figure 7.** Figure 7: Examples of manipulation scenarios of physical experiments. The [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

27 extracted references · 18 canonical work pages

[7]

Learning to grasp the ungraspable with emer- gent extrinsic dexterity,

W. Zhou and D. Held, “Learning to grasp the ungraspable with emer- gent extrinsic dexterity,” in Conference on Robot Learning . PMLR, 2023, pp. 150–160

work page 2023
[10]

Learning extrinsic dexterity with parameterized manipulation primitives,

S.-M. Yang, M. Magnusson, J. A. Stork, and T. Stoyanov, “Learning extrinsic dexterity with parameterized manipulation primitives,” in 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 5404–5410

work page 2024
[1]

Learning insertion primitives with discrete-continuous hybrid action space for robotic assembly tasks,

X. Zhang, S. Jin, C. Wang, X. Zhu, and M. Tomizuka, “Learning insertion primitives with discrete-continuous hybrid action space for robotic assembly tasks,” in 2022 International conference on robotics and automation (ICRA) . IEEE, 2022, pp. 9881–9887

work page 2022
[2]

Efficient sim-to-real transfer of contact-rich manipulation skills with online admittance residual learning,

X. Zhang, C. Wang, L. Sun, Z. Wu, X. Zhu, and M. Tomizuka, “Efficient sim-to-real transfer of contact-rich manipulation skills with online admittance residual learning,” in Conference on Robot Learn- ing. PMLR, 2023, pp. 1621–1639

work page 2023
[3]

Robotic object insertion with a soft wrist through sim-to-real privi- leged training,

Y . Fuchioka, C. C. Beltran-Hernandez, H. Nguyen, and M. Hamaya, “Robotic object insertion with a soft wrist through sim-to-real privi- leged training,” in 2024 IEEE/RSJ International Conference on Intel- ligent Robots and Systems (IROS) . IEEE, 2024, pp. 9159–9166

work page 2024
[4]

Deep reinforcement learning for high precision assembly tasks,

T. Inoue, G. De Magistris, A. Munawar, T. Yokoya, and R. Tachibana, “Deep reinforcement learning for high precision assembly tasks,” in 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2017, pp. 819–825

work page 2017
[5]

Manipulation Trajectory Optimization with Online Grasp Synthesis and Selection

L. Wang, Y . Xiang, and D. Fox, “Manipulation trajectory opti- mization with online grasp synthesis and selection,” arXiv preprint arXiv:1911.10280, 2019

work page Pith review arXiv 1911
[6]

Robust in-hand manipulation with extrinsic contacts,

B. Liang, K. Ota, M. Tomizuka, and D. K. Jha, “Robust in-hand manipulation with extrinsic contacts,” in 2024 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2024, pp. 6544–6550

work page 2024

Show all 27 references

[8]

Multi-stage re- inforcement learning for non-prehensile manipulation,

D. Wang, C. Liu, F. Chang, H. Huan, and K. Cheng, “Multi-stage re- inforcement learning for non-prehensile manipulation,” IEEE Robotics and Automation Letters , 2024

2024
[9]

Learning generalizable pivoting skills,

X. Zhang, S. Jain, B. Huang, M. Tomizuka, and D. Romeres, “Learning generalizable pivoting skills,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2023, pp. 5865–5871

2023
[11]

Extrinsic dexterity: In-hand manipulation with external forces,

N. C. Dafle, A. Rodriguez, R. Paolini, B. Tang, S. S. Srinivasa, M. Erdmann, M. T. Mason, I. Lundberg, H. Staab, and T. Fuhlbrigge, “Extrinsic dexterity: In-hand manipulation with external forces,” in 2014 IEEE International Conference on Robotics and Automation (ICRA). IEEE, ...

2014
[12]

Contact mode guided motion planning for quasidynamic dexterous manipulation in 3d,

X. Cheng, E. Huang, Y . Hou, and M. T. Mason, “Contact mode guided motion planning for quasidynamic dexterous manipulation in 3d,” in 2022 International Conference on Robotics and Automation (ICRA) . IEEE, 2022, pp. 2730–2736

2022
[13]

Object manipulation through contact configuration regulation: multiple and intermittent contacts,

O. Taylor, N. Doshi, and A. Rodriguez, “Object manipulation through contact configuration regulation: multiple and intermittent contacts,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2023, pp. 8735–8743

2023
[14]

Stocs: Simultaneous trajectory optimization and contact selection for contact- rich manipulation,

M. Zhang, D. K. Jha, A. Raghunathan, and K. Hauser, “Stocs: Simultaneous trajectory optimization and contact selection for contact- rich manipulation,” in Embracing Contacts-Workshop at ICRA 2023 , 2023

2023
[15]

One-shot transfer of long-horizon extrinsic manipulation through contact retargeting,

A. Wu, R. Wang, S. Chen, C. Eppner, and C. K. Liu, “One-shot transfer of long-horizon extrinsic manipulation through contact retargeting,” in 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2024, pp. 13 891–13 898

2024
[16]

Learning pregrasp manipulation of objects from ungraspable poses,

Z. Sun, K. Yuan, W. Hu, C. Yang, and Z. Li, “Learning pregrasp manipulation of objects from ungraspable poses,” in 2020 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2020, pp. 9917–9923

2020
[17]

Combo- grasp: Learning constraint-based manipulation for bimanual occluded grasping,

J. Yamada, A. L. Mitchell, J. Collins, and I. Posner, “Combo- grasp: Learning constraint-based manipulation for bimanual occluded grasping,” arXiv preprint arXiv:2502.08054 , 2025

2025 arXiv
[18]

6-dof graspnet: Variational grasp generation for object manipulation,

A. Mousavian, C. Eppner, and D. Fox, “6-dof graspnet: Variational grasp generation for object manipulation,” in Proceedings of the IEEE/CVF international conference on computer vision , 2019, pp. 2901–2910

2019
[19]

Learning diverse and physically feasible dexterous grasps with generative model and bilevel optimization,

A. Wu, M. Guo, and K. Liu, “Learning diverse and physically feasible dexterous grasps with generative model and bilevel optimization,” in Proceedings of The 6th Conference on Robot Learning , ser. Proceedings of Machine Learning Research, K. Liu, D. Kulic, and J. Ichnowski, Ed...

2023
[20]

Multi-fingered robotic hand grasping in cluttered environ- ments through hand-object contact semantic mapping,

L. Zhang, K. Bai, G. Huang, Z. Bing, Z. Chen, A. Knoll, and J. Zhang, “Multi-fingered robotic hand grasping in cluttered environ- ments through hand-object contact semantic mapping,” arXiv preprint arXiv:2404.08844, 2024

2024 arXiv
[21]

Decomposed vector-quantized variational autoencoder for human grasp generation,

Z. Zhao, M. Qi, and H. Ma, “Decomposed vector-quantized variational autoencoder for human grasp generation,” in European Conference on Computer Vision. Springer, 2024, pp. 447–463

2024
[22]

Hacman: Learning hybrid actor-critic maps for 6d non-prehensile manipulation,

W. Zhou, B. Jiang, F. Yang, C. Paxton, and D. Held, “Hacman: Learning hybrid actor-critic maps for 6d non-prehensile manipulation,” arXiv preprint arXiv:2305.03942 , 2023

2023 arXiv
[23]

Hac- man++: Spatially-grounded motion primitives for manipulation,

B. Jiang, Y . Wu, W. Zhou, C. Paxton, and D. Held, “Hac- man++: Spatially-grounded motion primitives for manipulation,” arXiv preprint arXiv:2407.08585, 2024

2024 arXiv
[24]

One-shot manipulation strategy learning by making contact analo- gies,

Y . Liu, J. Mao, J. Tenenbaum, T. Lozano-P ´erez, and L. P. Kaelbling, “One-shot manipulation strategy learning by making contact analo- gies,” arXiv preprint arXiv:2411.09627 , 2024

2024 arXiv
[25]

Playing atari with deep reinforcement learning,

V . Mnih, K. Kavukcuoglu, D. Silver, A. Graves, I. Antonoglou, D. Wierstra, and M. Riedmiller, “Playing atari with deep reinforcement learning,” arXiv preprint arXiv:1312.5602 , 2013

2013 arXiv
[26]

Soft actor-critic algorithms and applications,

T. Haarnoja, A. Zhou, K. Hartikainen, G. Tucker, S. Ha, J. Tan, V . Kumar, H. Zhu, A. Gupta, P. Abbeel et al. , “Soft actor-critic algorithms and applications,” arXiv preprint arXiv:1812.05905 , 2018

2018 arXiv
[27]

Apriltag: A robust and flexible visual fiducial system,

E. Olson, “Apriltag: A robust and flexible visual fiducial system,” in 2011 IEEE international conference on robotics and automation . IEEE, 2011, pp. 3400–3407

2011

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [7]

Learning to grasp the ungraspable with emer- gent extrinsic dexterity,

W. Zhou and D. Held, “Learning to grasp the ungraspable with emer- gent extrinsic dexterity,” in Conference on Robot Learning . PMLR, 2023, pp. 150–160

work page 2023

[2] [10]

Learning extrinsic dexterity with parameterized manipulation primitives,

S.-M. Yang, M. Magnusson, J. A. Stork, and T. Stoyanov, “Learning extrinsic dexterity with parameterized manipulation primitives,” in 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 5404–5410

work page 2024

[3] [1]

Learning insertion primitives with discrete-continuous hybrid action space for robotic assembly tasks,

X. Zhang, S. Jin, C. Wang, X. Zhu, and M. Tomizuka, “Learning insertion primitives with discrete-continuous hybrid action space for robotic assembly tasks,” in 2022 International conference on robotics and automation (ICRA) . IEEE, 2022, pp. 9881–9887

work page 2022

[4] [2]

Efficient sim-to-real transfer of contact-rich manipulation skills with online admittance residual learning,

X. Zhang, C. Wang, L. Sun, Z. Wu, X. Zhu, and M. Tomizuka, “Efficient sim-to-real transfer of contact-rich manipulation skills with online admittance residual learning,” in Conference on Robot Learn- ing. PMLR, 2023, pp. 1621–1639

work page 2023

[5] [3]

Robotic object insertion with a soft wrist through sim-to-real privi- leged training,

Y . Fuchioka, C. C. Beltran-Hernandez, H. Nguyen, and M. Hamaya, “Robotic object insertion with a soft wrist through sim-to-real privi- leged training,” in 2024 IEEE/RSJ International Conference on Intel- ligent Robots and Systems (IROS) . IEEE, 2024, pp. 9159–9166

work page 2024

[6] [4]

Deep reinforcement learning for high precision assembly tasks,

T. Inoue, G. De Magistris, A. Munawar, T. Yokoya, and R. Tachibana, “Deep reinforcement learning for high precision assembly tasks,” in 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2017, pp. 819–825

work page 2017

[7] [5]

Manipulation Trajectory Optimization with Online Grasp Synthesis and Selection

L. Wang, Y . Xiang, and D. Fox, “Manipulation trajectory opti- mization with online grasp synthesis and selection,” arXiv preprint arXiv:1911.10280, 2019

work page Pith review arXiv 1911

[8] [6]

Robust in-hand manipulation with extrinsic contacts,

B. Liang, K. Ota, M. Tomizuka, and D. K. Jha, “Robust in-hand manipulation with extrinsic contacts,” in 2024 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2024, pp. 6544–6550

work page 2024