IROS 2026 project page

Probe-to-Grasp Manipulation Using Self-Sensing Pneumatic Variable-Stiffness Joints

Ngoc-Duy Tran1, Yeman Fan1, Feng Dai1, Khang Nguyen2, Anh Nguyen3, Hoang-Hiep Ly4, Tung D. Ta1,5, Shigeru Chiba1,

1The University of Tokyo, Japan

2Mohamed bin Zayed University of Artificial Intelligence, UAE

3University of Liverpool, UK

4Hanoi University of Science and Technology, Vietnam

5Keio University, Japan

A passive soft-rigid hybrid gripper uses pressure changes inside pneumatic soft-ring joints to probe local object stiffness, then selects stable grasp locations without embedded tactile arrays.

Paper Results Video
Hybrid gripper probing and grasping banana and orange objects
The hybrid gripper combines rigid links with pneumatic soft-ring joints for compliant probing and stiffness-aware grasping.

Overview

Abstract

Grasping deformable objects with varying stiffness remains a significant challenge in robotics. Estimating the local stiffness of a target object is important for determining an optimal grasp pose that enables stable pickup without damaging the object. This paper presents a probe-to-grasp manipulation framework for estimating the relative stiffness of objects using a passive soft-rigid two-finger hybrid gripper equipped with self-sensing pneumatic variable-stiffness joints.

Each finger consists of two rigid links connected by a soft pneumatic ring placed at the joint, enabling compliant interaction and controllable joint stiffness via internal pressurization. By measuring pressure inside the pneumatic ring, the system estimates interaction force during contact. The probing strategy then infers relative object stiffness by correlating estimated normal force with known gripper closing displacement.

01

Passive pneumatic hybrid gripper

Rigid finger links and soft-ring joints support both probing and grasping with a simple pneumatic system.

02

Pressure-only stiffness cue

Pressure variation in the soft ring is used as a low-cost signal for contact force and relative stiffness estimation.

03

Stiffness-aware grasp selection

Fruit experiments show how the gripper probes candidate regions and avoids softer locations during grasp planning.

Demonstration

Project Video

Video demonstration of the probe-to-grasp manipulation system.

Hardware

Self-Sensing Hybrid Gripper Design

Hardware details of the hybrid gripper finger and pneumatic soft-ring system
Structural components and dimensions of the finger, soft pneumatic ring, solenoid valve, and pressure sensor.

Variable-stiffness soft-ring joint

Each gripper finger uses two rigid links joined by a passive pneumatic soft ring. Internal pressure adjusts joint stiffness while preserving compliance during contact with fragile and deformable objects.

A pressure sensor monitors the pneumatic ring directly. When the finger contacts an object, bending and constrained air volume create a measurable pressure change that is mapped to interaction force through experimental characterization.

  • Two-finger soft-rigid hybrid gripper for probing and grasping.
  • Soft ring around the joint acts as both compliant element and sensing chamber.
  • Solenoid valves constrain air volume during probing to expose pressure variation.
Experimental platform with xArm 7 robotic arm, pressure regulator, control board, and hybrid gripper
Experimental platform with an xArm 7 robot arm, hybrid gripper, pressure regulator, pneumatic supply, and control board.

Probe-to-grasp

Pressure-Based Stiffness Estimation

1

Set initial pressure

The soft-ring joint is pressurized, then air volume is constrained by the pneumatic system.

2

Probe candidate contacts

The robot closes the gripper by known displacements while pressure variation is recorded.

3

Infer relative stiffness

For the same closing motion, stiffer contact regions produce larger pressure changes.

4

Select grasp location

The gripper chooses firmer regions for stable grasping and avoids soft areas that may be damaged.

ko = F / δ

Object stiffness is treated as force over deformation; pressure change provides the practical force cue.

ΔP ↑ → krelative ↑

Measured pressure variation increases with relative object stiffness under matched probing conditions.

Experiments

Characterization and Probing Results

450 N·mm maximum joint torque near 80° bending
0-80 kPa initial pressure range for constrained-air probing
60 kPa pressure condition with strong stiffness discrimination
Stiffness characterization procedure with force gauge and hybrid finger
Data-collection setup for validating stiffness estimation using a force gauge and controlled fingertip bending.
Torque map across pressure and bending angle
Joint torque rises with internal pressure and bending angle, reaching roughly 450 N·mm at high bending.
Torque map under constrained air volume
Under constrained air volume, torque still follows pressure and bending angle trends across 0-80 kPa.
Torque and pressure variation during increasing and decreasing bending phases
Pressure variation tracks bending and torque across increasing and decreasing phases at 60 kPa initial pressure.
Pressure variation during probing across different initial pressures and cube stiffnesses
Probing results for Cube 20A, Cube 30A, and Cube 80A show pressure change increasing with closing distance and relative stiffness.

Real-world validation

Stiffness-Aware Fruit Grasping

Banana stiffness probing result with possible and unsuitable grasping locations
Along the banana body, softer regions coincide with lower pressure change and are marked as unsuitable grasping locations.
Orange stiffness probing result around different angles
Around the orange, the gripper probes multiple angles and selects higher-stiffness regions for a stable grip.

Limitations and Future Work

The current system is limited by pressure sensor resolution, hysteresis in the soft material, air leakage at pneumatic fittings, and sensing on only one fingertip. Future work will use simulation and improved sensing hardware to refine the soft-ring design and improve repeatability across broader object classes.

Citation

BibTeX

@misc{tran2026probetograsp,
  title={Probe-to-Grasp Manipulation Using Self-Sensing Pneumatic Variable-Stiffness Joints},
  author={Tran, Ngoc-Duy and Fan, Yeman and Dai, Feng and Nguyen, Khang and Nguyen, Anh and Ly, Hoang-Hiep and Ta, Tung D. and Chiba, Shigeru},
  year={2026}
}