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

TL;DR

The paper tackles in-hand slip in multifingered grasps under unknown disturbances by proposing a hybrid data-driven and model-based control pipeline. It detects slip with a multimodal tactile stack (piezoelectric sensing for fast slip cues and piezoresistive arrays for contact localization) and online updates the grasp model to compute a null-space internal-force direction within the grasp matrix $\mathbf{G}$, ensuring object-wrench preservation while increasing contact normal forces. The end-to-end sensing-to-command latency is theoretically $35$-$40$ ms, with measured slip-detection delays around $20$ ms in controlled tests; closed-loop stabilization is demonstrated under external perturbations, validating robustness and rapid reaction. Across two- and three-finger precision grasps, a single internal-force step arrests slip without disturbing the object wrench, highlighting the practical potential for sub-$50$ ms closed-loop stabilization in future setups with continuous streaming and tighter perception-to-control integration.

Abstract

We present a hybrid learning and model-based approach that adapts internal grasp forces to halt in-hand slip on a multifingered robotic gripper. A multimodal tactile stack combines piezoelectric (PzE) sensing for fast slip cues with piezoresistive (PzR) arrays for contact localization, enabling online construction of the grasp matrix. Upon slip, we update internal forces computed in the null space of the grasp via a quadratic program that preserves the object wrench while enforcing actuation limits. The pipeline yields a theoretical sensing-to-command latency of 35-40 ms, with 5 ms for PzR-based contact and geometry updates and about 4 ms for the quadratic program solve. In controlled trials, slip onset is detected at 20ms. We demonstrate closed-loop stabilization on multifingered grasps under external perturbations. Augmenting efficient analytic force control with learned tactile cues yields both robustness and rapid reactions, as confirmed in our end-to-end evaluation. Measured delays are dominated by the experimental data path rather than actual computation. The analysis outlines a clear route to sub-50 ms closed-loop stabilization.

Figures (6)

• Figure 1: Modular 3-phalange finger with hybrid tactile pads. Each phalanx pairs spatial localization (PzR) with fast dynamics sensing (PzE) for the RSC pipeline.
• Figure 2: Overview of the Reactive Slip Control (RSC) pipeline, which integrates both sensing modalities of the PzE and PzR hybrid sensors. Piezoelectric sensors capture friction vibrations to detect slippage through spectral features and machine learning. Piezoresistive arrays provide data on pressure distribution and contact areas across the fingerpads. Using robot kinematics, this information helps construct the grasp matrix and identify a basis for internal forces within its null space. When slippage is detected, the grasp effort is adjusted by applying carefully selected force ratios to maintain object equilibrium.
• Figure 3: Slip control scenario with predefined force trajectories illustrating slip–detection and slip-control dynamics. Pull force increases from 3 N to 7 N, generating object motion. After a predetermined delay of 350ms, the grasp force is progressively augmented from 8 N to 16 N over 150 ms, providing sufficient friction to halt the slip. Detection delays are 30ms for slip onset and 130ms for slip offset. In this recording, the FFT visualization does not clearly isolate slip-specific features, but the different phases are distinguishable: (i) baseline vibrations from the robot arm, (ii) loss of adherence during slip (reduced high-frequency content), (iii) low-frequency dynamics introduced by force increase and surface deformation, and (iv) return to a stable spectral signature once adherence is restored.
• Figure 4: Slip control with symmetrical grasp by 2 parallel fingers. Reactive slip control is applied to stop detected slippage. PzR sensor show increasing contact pressure. Equilibrium is trivially obtained by applying equal forces in both fingers.
• Figure 5: Slip control with asymmetrical planar grasp using 3 fingers. An external traction force of 10 N is applied to the object generating slippage. Reactive slip control is performed after 130 ms, during which the object traveled 19 mm. The hybrid tactile sensor is used to detect slippage through piezoelectric sensing of friction vibrations. Grip force is increased to stop the slippage. The ratio of internal forces, ensuring the kinetostatic equilibrium, is computed in the kernel of the grasp matrix. To this end, piezoresistive sensors are used to estimate the location of contact points.