TacDexGrasp: Compliant and Robust Dexterous Grasping with Tactile Feedback

Yubin Ke; Jiayi Chen; Hang Lv; Xiao Zhou; He Wang

cs.ROMar 7, 2026

TacDexGrasp: Compliant and Robust Dexterous Grasping with Tactile Feedback

Authors: Yubin Ke, Jiayi Chen, Hang Lv, Xiao Zhou, He Wang

Organizations: Peking University · Galbot · Beijing Academy of Artificial Intelligence

Abstract

Multi-fingered hands offer great potential for compliant and robust grasping of unknown objects, yet their high-dimensional force control presents a significant challenge. This work addresses two key problems: (1) distributing forces across multiple contacts to counteract an object's weight, and (2) preventing rotational slip caused by gravitational torque when a grasp is distant from the object's center of mass. We address these challenges via tactile feedback and a Second-Order Cone Programming (SOCP)-based controller, without explicit torque modeling or slip detection. Our key insights are (1) rotational slip inevitably induces translational slip at some contact points for a multi-fingered grasp, and (2) the ratio of tangential to normal force at each contact is an effective early stability indicator. By actively constraining this ratio for each finger below the estimated friction coefficient, our controller maintains grasp stability against both translational and rotational slip. Real-world experiments on 12 diverse objects demonstrate the robustness and compliance of our approach.

Figures & tables

Fig. 1: TacDexGrasp. (A-C) Our system enables stable and safe grasping for multi-fingered hands on objects with diverse, unknown mass distributions, friction coefficients, and deformation materials. (D) Our SOCP-based controller can compensate for gravitational torque to prevent the rotational slip without explicit torque modeling.

Fig. 2: Illustration of our geometric insight: rotational slip inevitably induces translational slip. If the object (black) rotates about the $x$ -axis to the dashed pose (green), each $O_{i}$ moves to $O_{i}^{\text{new}}$ , while $O_{1}$ stays fixed because it is on the $x$ -axis.

Fig. 3: Real World Setup. (Left) 12 diverse objects. (Middle) Friction coefficients of each object. Note that both mass and friction coefficients are presented only as statistics and are not used in the experiments. (Right) The robotic platform used in our study.

	Baselines		Ablations		Ours
Object	COP	BODex	w/o PID	w/o SOCP	Ours
Rigid	2.01/70%	2.11/ 100%	1.57 /90%	1.58/60%	1.58/95%
Deformable	2.38/55%	3.62/70%	1.79/60%	2.20/35%	1.24 / 75%
Elongated	3.29/25%	3.56/45%	3.14/40%	2.83/20%	1.96 / 80%
Total	2.56/50%	3.10/72%	2.17/63%	2.20/38%	1.59 / 83%

TABLE I: Baseline comparison and ablation study. Each entry reports the relative max force

F_{max}/G

(lower is better) and success rate (higher is better). Our method achieves the best overall trade-off, particularly on deformable and elongated objects. Removing either the PID or SOCP module leads to notable degradation.

Fig. 4: Real-World Experiments. For diverse objects with varying masses, deformability, and friction, our system rapidly adapts and achieves stable grasps. The measured contact forces align closely with the SOCP-derived target forces. Legends denote fingertip labels: TH (thumb), LF (little finger), MF (middle finger), and FF (forefinger).

Category	Object	$\tilde{G}/G$	$\tilde{\mu}/\mu$	$F_{max}/G$	Success ( $\#$ /5)
Rigid	1-Apple	1.20	0.88	2.01	4/5
	2-Tissue	1.10	0.81	1.83	5/5
	3-Oreo Box	1.06	0.86	1.38	5/5
	4-White Box	0.98	0.60	1.11	5/5
Deformable	5-Soft Toy	0.82	1.11	1.03	5/5
	6-Half Tissue	1.11	0.90	1.35	3/5

TABLE II: Detailed quantitative results across object categories. Our method achieves high success rates while adaptively maintaining low contact forces across all objects.

Fig. 5: Mass Adaptation Test. Our system quickly responds to two sudden mass increases by increasing the target normal forces accordingly.

Fig. 6: Rapid Shaking Test. Our system accurately estimates object weight during vigorous shaking by compensating for acceleration-induced disturbances.

Fig. 7: Irregular Motion Test. The system holds a fragile crisp through irregular motions without causing damage.

Explore similar work

Reactive Slip Control in Multifingered Grasping: Hybrid Tactile Sensing and Internal-Force Optimization
Feb 18, 2026Théo Ayral, Saifeddine Aloui, Mathieu GrossardRobotic Grasping Contact-Rich Robotic Manipulation
Université Grenoble Alpes, CEA, Leti, F-38000 Grenoble, France · Université Paris-Saclay, CEA, List, F-91120 Palaiseau, France
Tracing Energy Flow: Learning Tactile-based Grasping Force Control to Reduce Slippage in Dynamic Object Interaction
Dec 24, 2025Cheng-Yu Kuo, Hirofumi Shin, Takamitsu MatsubaraRobotic Grasping Contact-Rich Robotic Manipulation
Graduated School of Science and Technology, Nara Institute of Science and Technology, Nara 630-0192, Japan · Honda R&D, Ltd., Saitama 351-0114, Japan
Temporal Visuo-Tactile Learning for Dexterous Grasp Stability
Oct 7, 2026Ken Nakahara, Aleksei Buvailik, Prokhor Kotov +1Robotic Grasping Tactile Sensing for Robotic Manipulation
LASR Lab, Faculty of Computer Science, TU Dresden, Germany

Feb 18, 2026J/K move · Enter open · S save

Reactive Slip Control in Multifingered Grasping: Hybrid Tactile Sensing and Internal-Force Optimization

Théo Ayral, Saifeddine Aloui, Mathieu Grossard

Université Grenoble Alpes, CEA, Leti, F-38000 Grenoble, France · Université Paris-Saclay, CEA, List, F-91120 Palaiseau, France

Robotic Grasping Contact-Rich Robotic Manipulation Tactile Sensing for Robotic Manipulation

Open paper AskPDF arXiv

We build a low-level reflex control layer driven by fast tactile feedback for multifinger grasp stabilization. Our hybrid approach combines learned tactile slip detection with model-based internal-force control to halt in-hand slip while preserving the object-level wrench. The multimodal tactile stack integrates piezoelectric sensing (PzE) for fast slip cues and piezoresistive arrays (PzR) for contact localization, enabling online construction of a contact-centric grasp representation without prior object knowledge. Experiments demonstrate reactive stabilization of multifingered grasps under external perturbations, without explicit friction models or direct force sensing. In controlled trials, slip onset is detected after 20.4 +/- 6 ms. The framework yields a theoretical grasp response latency on the order of 30 ms, with grasp-model updates in less than 5 ms and internal-force selection in about 4 ms. The analysis supports the feasibility of sub-50 ms tactile-driven grasp responses, aligned with human reflex baselines.

Abstract

Explore similar work

Reactive Slip Control in Multifingered Grasping: Hybrid Tactile Sensing and Internal-Force Optimization

Tracing Energy Flow: Learning Tactile-based Grasping Force Control to Reduce Slippage in Dynamic Object Interaction

Temporal Visuo-Tactile Learning for Dexterous Grasp Stability