Project Description

Concept of OmniFiber

Thin linear forms are a humble yet ubiquitous geometric building blocks found throughout nature and the human body. For example, human skeletal muscle tissues, which help navigate our locomotion and consciously controlled movements, are a bundle of muscle fiber modules that compose 30% of our body weight. Such primitive artificial muscle fibers resembling those in the human body have been explored in the field of robotics as flexible exoskeletons to assist body movements. In this work, we utilize such thin and flexible fiber form factor as a configurable building-block to create dynamic movement-based interactions.

We utilize OmniFiber as an enabling technology for designing interactions with dynamic motion using a fiber form factor. The design system is based on thin fluidic artificial muscles that consist of filaments wrapped around elastomeric chambers surrounded by a woven sleeve that mechanically alters the actuation behavior. The actuators are thin (ø = 600 μm to 1.8 mm) and flexible enough improving conformability of artificial muscles for wearables, while having fast response time (<50 Hz), and high force outputs (F→ 19N). 

Design Space and Functionality

The design space of OmniFibers can be divided into two: primitive fiber architecture and fiber compositions.

The multilayered design of OmniFiber constitutes an internal working fluid, resistive sensor-traced inner tubing, an interlocked outer mesh, and an inextensible mechanical constraint to program the overall behavior. The integrated soft sensor allows the fiber to capture strain and simple bending behavior in real time to provide immediate haptic and visual feedback.

An application may require a different basic OmniFiber mechanism which mainly consists of similar components. The contractible-type fiber is typically more useful for high force output applications which can be of use for movement assistance and rehabilitation wearables, whereas the extensible-type renders larger geometric deformation, e.g. elongation, coiling, for applications such as soft wearable displays and expressive robotic crafting applications.

Mechanical Programming Pipeline

With our selective strain limiting technique which we call "mechanical constraint", we demonstrate more complex morphing behaviors (e.g. bending, coiling) with OmniFibers beyond simple axial motions such as extension and contraction demonstrated by previous fluidic artificial muscle research.

We present two mechanical constraint techniques to program the morphing behavior:

•  an inlaid non-elastic filament technique to manipulate the behavior of the braided outer mesh
• 3D-printed flexible constraint that can be relocated to change the strain-limiting section as can be seen in the video below.

Higher Hierarchical Structures: Robotic Assemblies and Fabrics

One core feature of our design system is modularity. This feature enables the user to customize the scale, assembly and configuration of the closed-loop controlled fibers for a given application. For example, if an application requires higher force outputs, the fibers can be simply arrayed, bundled or woven in a fabric form.

We present fiber compositions all of which are fabricated on traditional textile manufacturing machinery such as a flat bed weft knitting machine. As OmniFiber’s polyester outer mesh provides suitable friction for machine-knitting, these techniques intend to demonstrate the scalability of OmniFiber to higher hierarchical structures. 

By using long (>30 m) active fibers (ø= 0.6 - 0.9 mm) we exemplify weaving, plain, rib and inlay knitting (Figure a-c). Additionally, we devised a method to fabricate a 3D quasi-spacer fabric (Figure d) on a weft knitting machine. The active ’spacer’ layer is entirely made of a contractible-type OmniFiber, resulting in out-of-plane haptic feedback once actuated.

Haptic Modalities

As OmniFiber can be easily embedded into clothing, it makes a viable material for interaction designers to fabricate a variety of on-skin and on-body haptic sensations such as pulling, squeezing, poking, vibrating, and skin stretching.

Multiple haptic properties can be combined in a single OmniFiber device, such as lateral skin stretch, compression, textured, and high frequency vibration feedback allowing for immediate response (axial displacement peaking at 150 mm/s), high strains (up to 245%) and high force output (up to 19 N). 

System Architecture

The overall design of our system consists of the following components: a flexible and stretchable fiber-based modular interface, a pneumatic development platform, with an accessible web-based GUI, and a multi-channel, multiplexed analog input for closed-loop strain control. 

Our system is also modular at the control level, using the FlowIO Platform, a miniaturized integrated development platform for driving soft pneumatically-actuated devices. 

We appropriated the FlowIO design to fit the following requirements of our system; yielding high pressures (0-400 kPa) and BLE communication with the multichannel resistive circuit allowing for closed-loop control and bidirectional interactions, and additional features for tangible programmability that can be found in our paper.

Closed-loop Strain Control

We use the FlowIO web app running in Chrome to receive the multichannel strain data from the 16-pin analog input module based on tangible deformation input by an OmniFiber device, process it in real-time, and control the output pressure (in the same OmniFiber device, or a paired second device) based on this data. 

Applications

In our demonstrations, we explored how OmniFiber-based devices can be utilized to develop novel user interactions; for example augmenting users’ natural muscles when playing an instrument, or recording and replaying the respiratory physiology of a classically trained singer.

We also implemented dynamic line-based interfaces for display and affordance such as interactive cables, soft on-body displays for sending visuo-haptic messages to a loved one, and robotic jewelry that adapts to its owner's body movements. 

For more details, please visit our UIST 2021 paper below.

By Ozgun Kilic Afsar, Ali Shtarbanov, Hila Mor, Ken Nakagaki, Jack Forman, Dr. Seung Hee Jeong, 
Prof. Klas Hjort, Prof. Kia Hook, Prof. Hiroshi Ishii. ACM Symposium on User Interface Software
and Technology (UIST'21), Virtual, October 2021. DOI:  https://doi.org/10.1145/3472749.3474802