IORA: Irregular Object Immobilization under Restricted Actuation

Sang-Hun Han; Changyeon Won; JunGyu Lee; Min-Gyu Park; Hae-Gon Jeon

cs.ROOct 22, 2026

IORA: Irregular Object Immobilization under Restricted Actuation

Authors: Sang-Hun Han, Changyeon Won, JunGyu Lee, Min-Gyu Park, Hae-Gon Jeon

Organizations: Gwangju Institute of Science and Technology · Yonsei University · Korea Electronics Technology Institute · Polygom

Abstract

Stable grasping under task-specific disturbances is important for robotic manipulation. However, most dexterous grasp synthesis methods are developed for highly actuated hands in free space, and low-DoF hands may not be able to realize the contact topologies required by these grasps. Meanwhile, support surfaces can provide useful contact forces, but are often treated only as geometric constraints. We present a support-aware grasp synthesis framework for object immobilization with low-DoF coupled hands. Given irregular object geometry, our method represents both hand and object local surface regions as graphs and searches for kinematically feasible matchings between them. We then refine the wrist pose and motor state for geometric contact feasibility and validates the refined grasps through physical rollouts. Starting from the validated grasp pose, an optional task-guided local search explores nearby grasp configurations by varying the wrist pose and motor state, and selects one according to the specified task condition. Experiments on diverse objects show that our method generates various grasp configurations that achieve high grasp success and strong disturbance resistance. Real-world experiments further demonstrate physically executable grasps which enable object immobilization in cooking scenario.

Figures & tables

Fig. 3: Qualitative comparison with BODex and Dexonomy on eight objects. For each grasp we apply external forces along the tested axis directions in MuJoCo and report the maximum force resisted along the weakest direction, capped at 50 N. Red arrows show this weakest direction for each grasp. Green solid circles mark bracing contacts where our grasps wrap the object with multiple links and press it against the support surface. Red dashed circles mark defective contacts of the baselines such as fingertip-only touch, flat fingers resting on the surface, and one-sided contact without opposition.


Setting	Method	Success		Resistance							Contact Quality
Setting	Method	GSR $\uparrow$	OSR $\uparrow$	$F_{\min}\uparrow$	R@5	R@10	R@20	R@30	R@40	R@50	PD $\downarrow$	SPD $\downarrow$	CLN $\uparrow$	CDC $\downarrow$
		(%)	(%)	(N)	(%)	(%)	(%)	(%)	(%)	(%)	(mm, avg./med.)	(mm)		(mm, med.)

Tabletop	Ours	77.7	88.5	10.9	68.1	53.9	37.2	27.5	22.5	18.6	1.01 / 0.09	0.00	3.03	0.28
	Dexonomy	80.2	42.5	3.6	37.4	18.2	2.7	1.6	0.5	0.5	0.04 / 0.01	0.00	3.02	0.25

TABLE I: Quantitative comparison on the Low-DoF RH56 under floating and tabletop settings.


Scene	Method	Success		Resistance							Contact Quality
Scene	Method	GSR $\uparrow$	OSR $\uparrow$	$F_{\min}\uparrow$	R@5	R@10	R@20	R@30	R@40	R@50	PD $\downarrow$	SPD $\downarrow$	CLN $\uparrow$	CDC $\downarrow$
		(%)	(%)	(N)	(%)	(%)	(%)	(%)	(%)	(%)	(mm, avg./med.)	(mm)		(mm, med.)

Tabletop	Ours	88.1	93.5	9.8	64.1	49.6	34.0	22.2	16.6	11.9	0.83 / 0.41	0.06	3.96	0.07
	Dexonomy	86.0	95.0	5.7	54.8	24.0	7.2	1.5	0.2	0.2	1.18 / 0.87	0.02	5.12	0.02

TABLE II: Quantitative comparison on the High-DoF Shadow Hand under the no-selection protocol.


Variant	GSR	OSR	$F_{\min}$	R@5/10/20	PD avg./med.	CLN
		(%)	(%)	(N)	(%)	(mm)

Graph-based configuration	77.8	93.0	16.0	70.2/61.2/45.0	4.07/2.71	4.52
+ Geometric refinement	60.5	79.5	5.3	50.2/40.8/28.2	3.15/0.16	3.24
+ Physics validation	86.3	93.5	11.0	71.5/53.1/31.9	1.01/0.09	3.07

TABLE III: Analysis of the grasp synthesis pipeline on the coupled RH56 in the tabletop setting.


Metric	With support	Without support

Force-balance feasible (%)	78.9	17.9
Median predicted resistance (N)	13.3	0.0

TABLE IV: Static SOCP force-balance analysis of settled grasps with tabletop contacts included or removed.


Objective	GSR $\uparrow$	OSR $\uparrow$	$F_{\min}\uparrow$	R@5/10/20/30 $\uparrow$
	(%)	(%)	(N)	(%)

A: Geometry only	77.7	88.5	10.9	68.1 / 53.9 / 37.2 / 27.5
B: Calibrated SOCP	66.8	76.5	8.1	58.7 / 45.3 / 29.8 / 22.9
C: Diff. task-GWS ( $\pm 12$ mm)	73.2	80.5	10.8	61.4 / 52.8 / 37.5 / 28.9

TABLE V: Objective ablation for region-constrained geometric refinement on the RH56 under a shared evaluation protocol.


Condition	Median	R@5	R@10	R@15	R@20	R@25	R@30
	(N)	(%)	(%)	(%)	(%)	(%)	(%)

Base	10.7	67.2	52.3	41.4	35.3	29.9	25.5
$+x$ conditioned	16.9	73.5	63.0	51.7	45.8	41.2	34.0
$-x$ conditioned	17.9	73.1	61.3	52.9	46.6	38.2	33.2

TABLE VI: Direction-conditioned resistance after local derivation. Resistance is evaluated along each horizontal direction.


Squeeze coefficient	0.1	0.3	0.5	0.8	1.2

$F_{\min}$ median (N)	8.0	10.0	12.4	14.1	16.5
6-dir. success (%)	86	88	89	89	86
Settled penetration median (mm)	0.09	0.09	0.09	0.09	0.09

TABLE VII: Effect of squeeze strength on resistance and settled penetration for fixed RH56 grasps.

Fig. 4: Qualitative results of grasping real objects. The center image displays real objects used in the experiments and the surrounding images illustrate diverse grasp configurations adapting to each object shape.

Fig. 5: Experimental setup for evaluating disturbance resistance in the real world. The yellow arrow denotes the disturbance force. This force is caused by the gravity of the weights and transmitted to the objects via a pulley.


Object	GSR	R@10	R@15	R@20	R@25	R@30

1	100	100	100	66.7	33.3	33.3
2	100	100	100	66.7	33.3	33.3
3	66.7	66.7	0.0	0.0	0.0	0.0
4	100	100	66.7	66.7	66.7	66.7

TABLE VIII: Real-world evaluation of grasp success and disturbance resistance.

R@F

denotes the success rate under forces

F

Fig. 6: Demonstration of real-world deployment in a cooking scenario. Our synthesized grasps demonstrate high stability against the forces caused by cutting or peeling vegetables.

Robotic Grasping

Robotic grasping studies how robots perceive, plan, and execute stable, task-appropriate contacts with objects across varied shapes and conditions.

See all 316 papers

Adversarial Game-Theoretic Algorithm for Dexterous Grasp Synthesis
Nov 8, 2025Yu Chen, Botao He, Yuemin Mao +7Robotic Grasping Dexterous Manipulation
Robotics Institute, Carnegie Mellon University, PA 15213 · Department of Computer Science, University of Maryland, MD 20742 · Department of Computer and Information Science, University of Pennsylvania, PA 19104
Real-Time Force Regulation for Whole-Hand Dexterous Grasping
Sep 24, 2026Sang Min Kim, Alexander Alexiev, Tzu-Yuan Lin +3Robotic Grasping Dexterous Manipulation
Department of Electrical and Computer Engineering, Seoul National University, Seoul, Republic of Korea. · Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA. · Department of Artificial Intelligence, Yonsei University, Seoul, Republic of Korea.
Grasp to Act: Dexterous Grasping for Tool Use in Dynamic Settings
Feb 24, 2026Harsh Gupta, Mohammad Amin Mirzaee, Wenzhen YuanRobotic Grasping Dexterous Manipulation
University of Illinois Urbana-Champaign

Nov 8, 2025J/K move · Enter open · S save

Adversarial Game-Theoretic Algorithm for Dexterous Grasp Synthesis

Yu Chen, Botao He, Yuemin Mao, Arthur Jakobsson+6

Robotics Institute, Carnegie Mellon University, PA 15213 · Department of Computer Science, University of Maryland, MD 20742 · Department of Computer and Information Science, University of Pennsylvania, PA 19104

Robotic Grasping Dexterous Manipulation Robotic Manipulation

Open paper AskPDF arXiv

For many complex tasks, multi-finger robot hands are poised to revolutionize how we interact with the world, but reliably grasping objects remains a significant challenge. We focus on the problem of synthesizing grasps for multi-finger robot hands that, given a target object's geometry and pose, computes a hand configuration. Existing approaches often struggle to produce reliable grasps that sufficiently constrain object motion, leading to instability under disturbances and failed grasps. A key reason is that during grasp generation, they typically focus on resisting a single wrench, while ignoring the object's potential for adversarial movements, such as escaping. We propose a new grasp-synthesis approach that explicitly captures and leverages the adversarial object motion in grasp generation by formulating the problem as a two-player game. One player controls the robot to generate feasible grasp configurations, while the other adversarially controls the object to seek motions that attempt to escape from the grasp. Simulation experiments on various robot platforms and target objects show that our approach achieves a success rate of 75.78%, up to 19.61% higher than the state-of-the-art baseline. The two-player game mechanism improves the grasping success rate by 27.40% over the method without the game formulation. Our approach requires only 0.28-1.04 seconds on average to generate a grasp configuration, depending on the robot platform, making it suitable for real-world deployment. In real-world experiments, our approach achieves an average success rate of 85.0% on ShadowHand and 87.5% on LeapHand, which confirms its feasibility and effectiveness in real robot setups.

Abstract

Robotic Grasping

Explore similar work

Adversarial Game-Theoretic Algorithm for Dexterous Grasp Synthesis

Real-Time Force Regulation for Whole-Hand Dexterous Grasping

Grasp to Act: Dexterous Grasping for Tool Use in Dynamic Settings