RoboAtlas
EN
Festo logoFesto

BionicSoftHand

Price
Unknown
Type
Components · Dexterous Hands
Status
Prototype
Availability
Not released
Release Date
2019
Country/Region
Germany
Latest parameter verification
2026-10-30
5
Finger count

BionicSoftHand is a pneumatic soft robotic hand developed by Festo with air-chamber bellows fingers enclosed in a 3D textile of elastic and high-strength threads to direct expansion and motion. Festo’s research describes reinforcement learning in a digital-twin virtual environment for rotating a twelve-sided object until a specified face points upward, combining soft mechanics with learned control.

1 / 1BionicSoftHand

Technical specifications

Model specifications are shown below. Configuration-specific values are listed separately; missing data does not mean zero or unsupported.

Physical Specifications

Finger count
5
Actuator technology
Pneumatic
Frame Material
Housing made from 3D printing material
Materials
Flexible bellows and 3D textile coat
Materials
EPDM bellows, approximate Shore hardness 45
Materials
Silicone skin
Materials
Stereolithography resin airflow plate
Surface Fabric
Dyneema textile covering
Total degrees of freedom
12 DOF

Version-specific official brochure; v2 total includes wrist

Transmission mechanism
Pneumatic bellows
Actuator Model
Two EV pneumatic swivel modules for thumb and index lateral motion

Motion and coordination

Coordination architecture
Twelve VTEM piezo cartridges, twenty-four proportional piezo valves
Energy domain
Pneumatic energy
Hand Payload
Orientation dependent: ≤4 kg

Mechanical and mounting

Length
Each long finger: 9.8 cm
Length
Little finger: 7.9 cm
Mounting interface
Digital valve terminal mounted directly on the hand

Mechanical Structure

Dexterous Hand
BionicSoftHand dexterous hand
Dexterous Hand Model
BionicSoftHand
End-Effector
Pneumatic gripper
Joint Sensor Type
Ten inertial sensors in fingers

Sensors

Attitude Sensor
One inertial sensor on back of hand
Camera
One Leap Motion depth-sensing camera for computer vision
Depth Camera
Depth-sensing camera for computer vision
Sensors
Fourteen pressure sensors in airflow plate
Tactile Sensor
Fifteen tactile force sensors in fingers

Battery and Runtime

Power Supply
Compressed air: 3.5 bar for fingers, 6.0 bar for swivel modules

Compute Platform

AI Model
Reinforcement learning
Software Algorithm
Reinforcement learning in parallel digital-twin simulations, transferred to physical hand

Safety and Protection

Human-Robot Safety
Direct human-robot interaction
Safety System
Pneumatic operation for safe interaction

Interaction and Communication

External Port
One supply-air tube and one exhaust-air tube
Visual Recognition
Computer vision from a depth camera

Development and Expansion

Integration Interface
BionicSoftArm
Operating Modes
Goal-directed reinforcement learning

Configurations

BionicSoftHand

Sensing

Has tactile sensing
Supported

Commercial Status

Status
Prototype
Price
Unknown
Release Date
2019
Submit a source or correction